The 'Security Digest' Archives (TM)

Archive: About | Browse | Search | Contributions | Feedback
Site: Help | Index | Search | Contact | Notices | Changes

ARCHIVE: TCP-IP Distribution List - Archives (1994)
DOCUMENT: TCP-IP Distribution List for May 1994 (597 messages, 337364 bytes)
SOURCE: http://securitydigest.org/exec/display?f=tcp-ip/archive/1994/05.txt&t=text/plain
NOTICE: securitydigest.org recognises the rights of all third-party works.

START OF DOCUMENT

-----------[000000][next][prev][last][first]----------------------------------------------------
Date:      1 May 94 12:49:15 PDT
From:      burnhosp@vmsmail.gov.bc.ca
To:        comp.protocols.tcp-ip
Subject:   Meditech and TCP/IP

I am interested in hearing from anyone using Meditech 
over TCP/IP with Telnet. I have some specific and some
general questions and can offer some valuable information
that I've discovered during our own implementation. ANY
information or discussion would be welcome.
I can't believe we are alone in doing this...

Chris Hansen
Burnaby Hospital Society
(604) 431-4748
(604) 431-4802 (FAX)
burnhosp@galaxy.gov.bc.ca
-- 
Disclaimer: The opinions and statements contained in this posting are the sole
responsibility of the author and have not in any way been reviewed or approved
by my employer or any network service.

-----------[000001][next][prev][last][first]----------------------------------------------------
Date:      Sun,  1 May 1994 07:43:37 UTC
From:      an94584@anon.penet.fi
To:        comp.protocols.tcp-ip
Subject:   Multiple masks on a single machine



I have a question about running a SVR4 box with 3 interfaces ( ethernet
controllers)
The box will use a class B address aa.bb.0.0.
The ip address of the interfaces will be aa.bb.250.100
aa.bb.32.1 and aa.bb.128.1

If I setup the interfaces as follows:

IP ADDRESS              SUBNET MASK
== =======              ===========
aa.bb.250.100           FF.FF.00.00
aa.bb.128.1             FF.FF.80.00
aa.bb.32.1              FF.FF.E0.00

Will that allow me to access the aa.bb.128.0 sunet over the 128.0 interface
the 32.0 subnet over the 32.0 interface and the remainder over the 
250.0 interface

Thanks




-------------------------------------------------------------------------
To find out more about the anon service, send mail to help@anon.penet.fi.
Due to the double-blind, any mail replies to this message will be anonymized,
and an anonymous id will be allocated automatically. You have been warned.
Please report any problems, inappropriate use etc. to admin@anon.penet.fi.

-----------[000002][next][prev][last][first]----------------------------------------------------
Date:      2 May 1994 01:22:10 -0400
From:      davidpelo@aol.com (DavidPelo)
To:        comp.protocols.tcp-ip
Subject:   Re: DOS TCP/IP Stacks?

In article <43458@mindlink.bc.ca>, KCARPENT@mindlink.bc.ca (Ken Carpenter)
writes:

>I am looking for information on companies who offer TCP/IP stacks for DOS.
>
>Could anyone who knows of such products please email me with the >company's
>name, and if possible, the phone number or address.
>
>Thanks,

check with Microsoft -- it's FREE (see their download service or ftp from them)

-----------[000003][next][prev][last][first]----------------------------------------------------
Date:      Sun, 1 May 1994 19:38:52 GMT
From:      gnn@netcom.com (George Neville-Neil)
To:        comp.protocols.tcp-ip
Subject:   TCP/IP FAQ V1.1 June 1994

Hi Folks,

	I'm going to try and put out the FAQ on the first day of
every month from now on.  This version has some additions from 
the last as well as a table of contents.

Later,
George


Internet Protocol Frequently Asked Questions

Maintained by: George V. Neville-Neil (gnn@netcom.com)
Contributions from:
Stephane Bortzmeyer
Phill Conrad 
Jon Kay 
Barry Margolin 
W. Richard Stevens 
 
Version 1.1

Last Update:  May 1, 1994

************************************************************************

	The following is a list of Frequently Asked Questions, and
their answers, for people interested in the Internet Protocols,
including TCP, UDP, ICMP and others.  Please send all additions,
corrections, complaints and kudos to the above address.  This FAQ will
be posted on or about the first of every month.

************************************************************************

Table of Contents:
Glossary
1) Are there any good books on IP?
2) Where can I find example source code for TCP/UDP/IP?
3) Are there any public domain programs to check the performance of an
IP link? 
4) Where do I find RFCs?
5) How can I detect that the other end of a TCP connection has
crashed?  Can I use "keepalives" for this?
6) Can the keepalive timeouts be configured?
7) Can I set up a gateway to the Internet that translates IP
addresses, so that I don't have to change all our internal addresses 
to an official network? 


Glossary:

I felt this should be first given the plethora of acronyms used in the
rest of this FAQ.

IP: Internet Protocol.  The lowest layer protocol defined in TCP/IP.
This is the base layer on which all other protocols mentioned herein
are built.  IP is often referred to as TCP/IP as well.

UDP: User Datagram Protocol.  This is a connectionless protocol built
on top of IP.  It does not provide any guarantees on the ordering or
delivery of messages.  This protocol is layered on top of IP.

TCP: Transmission Control Protocol.  TCP is a connection oriented
protocol that guarantees that messages are delivered in the order in
which they were sent and that all messages are delivered.  If a TCP
connection cannot deliver a message it closes the connection and
informs the entity that created it.  This protocol is layered on top
of IP.

ICMP:  Internet Control Message Protocol.  ICMP is used for
diagnostics in the network.  The Unix program, ping, uses ICMP
messages to detect the status of other hosts in the net.

RFC: Request For Comment.  RFCs are documents that define the
protocols used in the IP Internet.  Some are only suggestions, some
are even jokes, and others are published standards.  Several sites in
the Internet store RFCs and make them available for anonymous ftp.

SLIP:  Serial Line IP.  An implementation of IP for use over a serial
link (modem).  CSLIP is an optimized (compressed) version of SLIP that
gives better throughput.

Bandwidth:  The amount of data that can be pushed through a link in
unit time.  Usually measured in bits or bytes per second.

Latency:  The amount of time that a message spends in a network going
from point A to point B.

Jitter:  The effect seen when latency is not a constant.  That is, if
messages experience a different latencies between two points in a
network.

RPC:  Remote Procedure Call.  RPC is a method of making network access
to resource transparent to the application programmer by supplying a
"stub" routine that is called in the same way as a regular procedure
call.  The stub actually performs the call across the network to
another computer.

Marshalling:  The process of taking arbitrary data (characters,
integers, structures) and packing them up for transmission across a
network.


1) Are there any good books on IP?

A) Yes.  Please see the following:

Internetworking with TCP/IP Volume I
(Principles, Protocols, and Architecture)
Douglas E. Comer
Prentice Hall 1991

This volume covers all of the protocols, including IP, UDP, TCP, and
the gateway protocols.  It also includes discussions of higher level
protocols such as FTP, TELNET, and NFS.

Internetworking with TCP/IP Volume II
(Design, Implementation, and Internals)
Douglas E. Comer / David L. Stevens
Prentice Hall 1991

Discusses the implementation of the protocols and gives numerous code
examples.

Internetworking with TCP/IP Volume III (BSD Socket Version)
(Client - Server Programming and Applications)
Douglas E. Comer / David L. Stevens
Prentice Hall 1993

This book discusses programming applications that use the internet
protocols.  It includes examples of telnet, ftp clients and servers.
Discusses RPC and XDR at length.

TCP/IP Illustrated, Volume 1: The Protocols, 
W. Richard Stevens
(c) Addison-Wesley, 1994 

An excellent introduction to the entire TCP/IP protocol suite,
covering all the major protocols, plus several important applications.

Unix Network Programming
W. Richard Stevens
Prentice Hall 1990

An excellent introduction to network programming under Unix.

The Design and Implementation of the 4.3 BSD Operating System
Samuel J. Leffler, Marshall Kirk McKusick, Michael J. Karels, John S.
Quarterman 
Addison-Wesley 1989

Though this book is a reference for the entire operating system, the
eleventh and twelfth chapters completely explain how the networking
protocols are implemented in the kernel.


2)  Where can I find example source code for TCP/UDP/IP?

A)  Code from the Internetworking with TCP/IP Volume III is available
for anonymous ftp from:

arthur.cs.purdue.edu:/pub/dls

Code used in the Net-2 version of Berkeley Unix is available for
anonymous ftp from:

ftp.uu.net:systems/unix/bsd-sources/sys/netinet 

and

gatekeeper.dec.com:/pub/BSD/net2/sys/netinet

Code from Richard Steven's book is available on:
ftp.uu.net:/published/books/stevens.*

3)  Are there any public domain programs to check the performance of
an IP link?

A)  

TTCP:  Available for anonymous ftp from....

Host gatekeeper.dec.com

    Location: /.0/BSD/NetBSD/NetBSD-current/othersrc
      DIRECTORY dr-xr-xr-x        512  Apr  8 09:57  ttcp
    Location: /.0/BSD/NetBSD/NetBSD-current/othersrc/ttcp
           FILE -r--r--r--       3885  Nov  7 03:35  ttcp.1
           FILE -r--r--r--      19225  Nov  7 03:35  ttcp.c

Host world.std.com

    Location: /src/wuarchive/graphics/graphics/mirrors/sgi.com/sgi/src/ttcp
           FILE -r--r--r--       3885  Oct  4 1991  ttcp.1
           FILE -r--r--r--      19170  May 17 1993  ttcp.c
           FILE -r--r--r--      13033  Sep  5 1989  ttcp.c-brl

There is suite of Bandwidth Measuring programs from gnn@netcom.com.
Available for anonymous ftp from ftp.netcom.com in
~ftp/gnn/bwmeas-0.3.tar.Z These are several programs that meausre
bandwidth and jitter over several kinds of IPC links, including TCP
and UDP.


4) Where do I find RFCs?

A)  RFCs are available for anonymous ftp from the following servers.
You should pick the one geographically closest to you. 

North America

FTP.NISC.SRI.COM , NIC.DDN.MIL and nic.cerf.net

Austrailia and Pacific Rim

munnari.oz.au

Denmark

ftp.diku.dk

Germany

walhalla.informatik.uni-dortmund.de

Finland

funet.fi

Netherlands

mcsun.eu.net

Norway

ugle.unit.no

Sweden

sunic.sunet.se and chalmers.se


Using Web, WAIS, and gopher:

Web:

http://web.nexor.co.uk/rfc-index/rfc-index-search-form.html

WAIS access by keyword:

wais://wais.cnam.fr/RFC

Excellent presentation with a full-text search too:

http://www.cis.ohio-state.edu/hypertext/information/rfc.html

With Gopher:

gopher://r2d2.jvnc.net/11/Internet%20Resources/RFC
gopher://muspin.gsfc.nasa.gov:4320/1g2go4%20ds.internic.net%2070%201%201/.ds/
.internetdocs



5) How can I detect that the other end of a TCP connection has crashed?
Can I use "keepalives" for this?

A) Detecting crashed systems over TCP/IP is difficult.  TCP doesn't require
any transmission over a connection if the application isn't sending
anything, and many of the media over which TCP/IP is used (e.g. ethernet)
don't provide a reliable way to determine whether a particular host is up.
If a server doesn't hear from a client, it could be because it has nothing
to say, some network between the server and client may be down, the server
or client's network interface may be disconnected, or the client may have
crashed.  Network failures are often temporary (a thin ethernet will appear
down while someone is adding a link to the daisy chain, and it often takes
a few minutes for new routes to stabilize when a router goes down), and TCP
connections shouldn't be dropped as a result.

Keepalives are a feature of the sockets API that requests that an empty
packet be sent periodically over an idle connection; this should evoke an
acknowledgement from the remote system if it is still up, a reset if it has
rebooted, and a timeout if it is down.  These are not normally sent until
the connection has been idle for a few hours.  The purpose isn't to detect
a crash immediately, but to keep unnecessary resources from being allocated
forever.

If more rapid detection of remote failures is required, this should be
implemented in the application protocol.  There is no standard mechanism
for this, but an example is requiring clients to send a "no-op" message
every minute or two.  An example protocol that uses this is X Display
Manager Control Protocol (XDMCP), part of the X Window System, Version 11;
the XDM server managing a session periodically sends a Sync command to the
display server, which should evoke an application-level response, and
resets the session if it doesn't get a response (this is actually an
example of a poor implementation, as a timeout can occur if another client
"grabs" the server for too long).

6) Can the keepalive timeouts be configured?

A) I know they can on many systems, but I don't know the details.

7) Can I set up a gateway to the Internet that translates IP addresses, so
that I don't have to change all our internal addresses to an official
network?

A) There's no general solution to this.  Many protocols include IP
addresses in the application-level data (FTP's "PORT" command is the most
notable), so it isn't simply a matter of translating addresses in the IP
header.  Also, if the network number(s) you're using match those assigned
to another organization, your gateway won't be able to communicate with
that organization (RFC 1597 proposes network numbers that are reserved for
private use, to avoid such conflicts, but if you're already using a
different network number this won't help you).

However, if you're willing to live with limited access to the Internet from
internal hosts, the "proxy" servers developed for firewalls can be used as
a substitute for an address-translating gateway. See the firewall FAQ.
-- 
gnn@netcom.com

Gentelmen, I will not have you fighting in the War Room.
					--- The President in Dr. Stragelove

-----------[000004][next][prev][last][first]----------------------------------------------------
Date:      Sun, 1 May 1994 19:39:39 GMT
From:      gnn@netcom.com (George Neville-Neil)
To:        comp.protocols.tcp-ip
Subject:   Whoops, that's the May FAQ.

Whoops.  The subject should have read May 1994.

Later,
George

-- 
gnn@netcom.com

Gentelmen, I will not have you fighting in the War Room.
					--- The President in Dr. Stragelove

-----------[000005][next][prev][last][first]----------------------------------------------------
Date:      Sun, 1 May 1994 20:43:56 GMT
From:      art@acc.com (Art Berggreen)
To:        comp.protocols.tcp-ip
Subject:   Re: Subnet mask

In article <Cp1ML9.EKE@tfs.com> ganesh@TFS.COM (Ganesh Vaidee) writes:
>Is it possible to have a subnet mask of 255.255.0.0 
>Does this mean that there are o bits for subnet id and 0 bits for host id.
>How is that possible. help. thanx

The first thing to understand, is that IP addresses and netmasks are
really 32 bit binary numbers.  The "dotted decimal" format that is so
familiar, is a notation convention that usually makes like easier,
but can complicate matters in some circumstances.  This format takes
the 32 bits, breaks them into four 8 bit chunks, and writes the decimal
value of each eight bit chunk, separated by periods.

A netmask is a 32 binary mask, which when applied to an IP address,
determines which bits of the address are "subnet part" and which
bits are "host part".  The 1 bits in the subnet mask denote the
subnet bits.  The world has come to expect the subnet part to be
the leftmost bits and be contiguous.

In the days before CIDR, subnet masks were usually equal to, or
wider that the "native netmask" which was determined by the IP
address class.  The address class was determined by the leftmost
bits of the address ( 0-class A, 10-class B, and 110-class C).
The native netmask for class A is 8 bits (255.0.0.0), for class B
is 16 bits (255.255.0.0), and for class C is 24 bits (255.255.255.0).
But it appears that IP address classes are being phased out, and
things will be described with IP address, netmask pairs.

Thus:

    subnet mask	255.255.0.0	11111111111111110000000000000000
    IP address  A.B.C.D		aaaaaaaabbbbbbbbccccccccdddddddd
				< subnet part  ><   host part  >

Note that subnet masks don't need to fall on eight bit boundaries,
but if they don't, dotted decimal representation is harder to work
with.

Hope this helps,
Art

-----------[000006][next][prev][last][first]----------------------------------------------------
Date:      1 May 1994 22:04:53 GMT
From:      dtix@max.tiac.net (Bud Biswas)
To:        comp.protocols.ppp,comp.protocols.tcp-ip,comp.protocols.tcp-ip.domains,comp.unix.admin,tamu.cs.general,tamu.networks,biz.sco.general
Subject:   Help on etherLAN setup with PPP, DNS

Hi,

I am trying to setup my ethernet LAN with internet connection from
internet provider via PPP link. I am having some problem in routing
PPP incoming traffic (from internet) to my entry machine's ethernet.
To give some background here is how my setup looks like:


						     ^ to internet
						     |
      port-server                                    |     Nameserver &
      	 |--------------|                     --------------- internet
         | lp1.tiac.net |                    | zork.tiac.net | gateway
         |  199.0.65.12 |                    | 199.0.65.2    |
         -------|--------                     -------|-------
                |				     |
                |            Subnet 199.0.65 (commercial internet provider)
      ==========|=================================================
                |
                |ppp link
                |
     ---------------------- 
     | alpha-ppp.dtix.com  |
     |     199.0.65.53     | 
     |---------------------|
     |   alpha.dtix.com    | 
     |   198.62.174.1      | 
     ----------------------  
                |
                |ethernet
                |            Subnet 198.62.174 (our own class C network)
      =============================================================


Versions of OS running on alpha.dtix.com is SCO SystemV version 3.2.4 with
TCP/IP runtime 1.2.1 with NET382 SLS applied. Our PPP link is working fine.
We are supposed to use internet provider's primary and secondary nameserver.
Here is what I am not able to configure.

(1) Configure DNS at our side, which is on alpha.dtix.com and will make use
    of internet provider's primary and secondary nameserver. They told me that
    I need to configure my nameserver as unauthoritative secondary nameserver.
    I would like to get some example configuration file like, named.soa, 
    named.hosts, named.local and named.boot which will help me in configuring
    my nameserver to make use of their primary & secondary nameserver. Also,
    what changes I need to make to any other relevant files on this machine
    or any other machine on my LAN. I read O'reilly's book on DNS and BIND,
    also another book Managing TCP/IP networks from O'reilly. But they
    discuss generally primary/secondary DNS setup in one domain. I couldn't
    find any example where two domains (dtix.com and tiac.net) are involved.
    I am sure someone out there must have gone through my situation.
    Any help please ??

(2) This question is for SCO sysadm's : I read in comp.protocols.ppp
    that SCO TCP/IP runtime has some bug which will not allow such
    co-existance of local ethernet with PPP connection and route traffic
    on appropriate channels. They have some patch for this. Has any body
    applied this patch? or does it work?

I would appreciate any help in this regard.

Thanks,

-Kalpesh Sheth
Digital Technology


-----------[000007][next][prev][last][first]----------------------------------------------------
Date:      1 May 1994 22:05:13 GMT
From:      barmar@think.com (Barry Margolin)
To:        comp.protocols.tcp-ip
Subject:   Re: What constitutes a duplicate TCP packet?

In article <cearley_k.33.000EC06D@wizard.colorado.edu> cearley_k@wizard.colorado.edu (Kent Cearley) writes:
>We have a problem with a host sending two tcp acknowledgements... one with 
>WIND=0 and no data, followed immediately with a WIND=256, also no data. Both 
>the sequence numbers and ack numbers are the same on both packets. The problem 
>is, a client (FTP Inc.'s WTN3270) sees the first packet, assumes it cannot 
>transmit, and appears to be discarding the second packet without the WIND=256 
>registering.
>
>FTP Inc. claims the RFC specifies if the SEQ# are the same its a duplicate 
>packet. If so, how is this situation handled? All of our other clients 
>recognize the second packet as advertising an available window. 

Since there's nothing in the TCP spec that says that duplicate packets
should be discarded without processing them, I don't see how this matters.
For instance, a duplicate data packet must be acknowledged, as it suggests
that the original acknowledgement was lost in the network.

The above is the normal way to open the TCP window on an idle connection.
You can't increase the sequence number if there's no data being sent.  What
does FTP Inc. expect your server to send?
-- 
Barry Margolin
System Manager, Thinking Machines Corp.

barmar@think.com          {uunet,harvard}!think!barmar

-----------[000008][next][prev][last][first]----------------------------------------------------
Date:      2 May 1994 08:38:17 -0600
From:      thayne@xmission.com (Thayne Forbes)
To:        comp.protocols.tcp-ip
Subject:   Re: Bargains on Comer's 3 Volume TCP/IP set

Cliff Bedore (cliffb@skeeter.bdb.com) wrote:
:> Scott Austin (scotta@kije.gsfc.nasa.gov) wrote:
:> : Hi,
 
:> : The Library of Computer and Information Sciences (LCIS), is currently
:> : having a promo with Comer's 3 books.  You can save some bucks, since the
:> : Publisher's prices on these is $164!  This offer is open to any
:> : non-members (wish *I* could).  The scoop is this:


:> The only problem is that I joined in March (maybe Feb) and I'm still waiting
:> for two of the Volumes

And I am still waiting for all three.  I have, on the other hand, got
three of the monthly mailers, and had to refuse the three monthly
selections.  (I am beginning to have reservations).

-- 
Thayne Forbes           thayne@xmission.com
Computer Weenie at large

-----------[000009][next][prev][last][first]----------------------------------------------------
Date:      1 May 1994 23:51:18 GMT
From:      jliv@wellfleet.com (John Loiacono IV)
To:        comp.protocols.tcp-ip,comp.dcom.sys.wellfleet,comp.dcom.sys.cisco
Subject:   Re: Variable IP subnets with RIP on Wellfleet and Cisco, routers

In article <2pjpq6$3tk@agate.berkeley.edu>
sklower@oboe.CS.Berkeley.EDU (Keith Sklower) writes:

> It would be good if somebody from wellfleet or
> cisco could authoritatively say whether their products support RIP version > 2.

We do not support RIPv2, at this time.

Regards,

John Loiacono IV
Wellfleet Communications, Inc.
jliv@wellfleet.com

-----------[000010][next][prev][last][first]----------------------------------------------------
Date:      Mon, 02 May 1994 13:24:09 -0500
From:      johns@oxygen.house.gov (John Schnizlein)
To:        comp.protocols.tcp-ip
Subject:   Re: Several gateways for two AS?

In article <16FAAE8A5.TCPMAINT@TWNMOE10.Edu.TW>, TCPMAINT@TWNMOE10.Edu.TW
wrote:

>     There are two Internet connection service providers in my country
> and we hope to establish several gateways in different locations to
> connect the two networks. Is it possible? Do we have to apply two
> different Aotonomous System Numbers for them? What routing protocol
> do we have to run on those gateways?

No!
You use one AS number and the Border Gateway Protocol (BGP) [RFC 1163] to
share route information with both Internet access providers. With BGP you
can establish policy as to which routes have precedence and what, if any,
transit traffic you will carry between your two bordering Autonomous
Systems.
-- 
Badges! we don't need no stinking badges!         |
disclaimer! we don't need no stinking disclaimer! | John M. Schnizlein
everybody knows nobody can represent the views of | johns@oxygen.house.gov
435 elected policy makers.                        | router jockey

-----------[000011][next][prev][last][first]----------------------------------------------------
Date:      2 May 94 12:48:39 PDT
From:      burnhosp@vmsmail.gov.bc.ca
To:        comp.protocols.tcp-ip
Subject:   Re: Bargains on Comer's 3 Volume TCP/IP set

I joined LCIS and had a little trouble with then in that
they sent me 2 books I had asked not to be sent. I returned them
and kept getting invoices until I sent them a letter explaining
the situation. I was cleared up very quickly after that.

Bottom line: I MIGHT join them again - I terminated my agreement
- but, like any mail order house, they must be closely monitored
and quick, detailed responses to errors must be sent.

In article <2q338p$f38@xmission.xmission.com>, thayne@xmission.com (Thayne Forbes) writes:
> Cliff Bedore (cliffb@skeeter.bdb.com) wrote:
> :> Scott Austin (scotta@kije.gsfc.nasa.gov) wrote:
> :> : Hi,
 
> :> : The Library of Computer and Information Sciences (LCIS), is currently
> :> : having a promo with Comer's 3 books.  You can save some bucks, since the
> :> : Publisher's prices on these is $164!  This offer is open to any
> :> : non-members (wish *I* could).  The scoop is this:
> 
> 
> :> The only problem is that I joined in March (maybe Feb) and I'm still waiting
> :> for two of the Volumes
> 
> And I am still waiting for all three.  I have, on the other hand, got
> three of the monthly mailers, and had to refuse the three monthly
> selections.  (I am beginning to have reservations).
> 
> -- 
> Thayne Forbes           thayne@xmission.com
> Computer Weenie at large
-- 
Disclaimer: The opinions and statements contained in this posting are the sole
responsibility of the author and have not in any way been reviewed or approved
by my employer or any network service.

-----------[000012][next][prev][last][first]----------------------------------------------------
Date:      2 MAY 94 11:45:14 EST
From:      ciarfella@took.enet.dec.com (Paul Ciarfella)
To:        comp.protocols.tcp-ip,comp.sys.mac.comm,comp.sys.novell
Subject:   Re: BootP on Novell's NLM and MacTCP


In article <opie-300494132746@opie.dialup.access.net>, opie@panix.com (Matthew S. "Opie" Warren) writes...
>What's up with this?  The only difference I can think of is that in the
>non-working site, the BootP server is in a different subnet from the
>requesting Macs.  Our networking people, however, assure me that this would
>not matter.

It does matter ... if the bootp server is in another subnet then you must
have a "relay agent" in between the subnets.  Your Wellfleet router should
be operating as a relay agent, which requires some special configuration
to be performed on the router..  

The relay agent knows the address of a bootp server and listens to bootp 
requests from your Macs.  When it receives the request, it should forward it 
directly to the bootp server.  The server will then respond back to the relay 
agent, which then sends the response (using the appropriate transport 
mechanism) to the requesting Mac.

Go back to the networking people and make sure that your routers have
the relay agent capability enabled.

Paul C

    ----------------------------------------------------------------------  
    Paul Ciarfella                             ciarfella@took.enet.dec.com
    Mobile & Wireless Networks                 508 486-7712
    Digital Equipment Corporation              Littleton, MA   USA
    ----------------------------------------------------------------------  





-----------[000013][next][prev][last][first]----------------------------------------------------
Date:      2 May 1994 20:05:50 -0700
From:      mshultz@crl.com (Michael Shultz)
To:        comp.protocols.tcp-ip
Subject:   Novell HELP!!!

I need some direction...

Here is the case. We have one network (Novell 3.12) that we have installed 
for billing, word processing etc... Another network running under Novell 
3.11, which we have a 56kbs dedicated Internet line comming into a 
CSU/DSU, then through a router and into an HP705 workstation, running 
HP/UX 9.07 and the server runs Netware NFS.
I have been thrown into this pot and I have plans of what I would like to 
do with the two of these networks but I am at a loss for direction...

Here are my plans, and if you can help at all, even in the slightest, It 
would be greatly appreciated!!!

I would like to use the 56kbs line for e-mail on the accounting network 
(we are currently running MSmail and AT&T Gateway), and also be able to 
telnet and/or ftp from certain workstations (all Windows3.1 workstations).
I would then like to have ftp and telnet on the other network (Lets call 
it the "Production" network, since it is the network that we receive ftp 
files from clients for printing on a Xerox Docutech and also on Scitex).

So! What software do I need, and what hardware do I need to make this all 
work smoothly? Big Question of the week...

If you can guide me in the right direction, and I can get this to work, 
there could be a reward in it for you! :^)

Anyway, any suggestions or guidence would be very appreciated...

Thank you in advance!

Mike Shultz
Lithocraft, Inc.
Santa Rosa, CA 95403
(707)544-8681


-----------[000014][next][prev][last][first]----------------------------------------------------
Date:      2 May 1994 08:44:45 GMT
From:      tli@cisco.com (Tony Li)
To:        comp.protocols.tcp-ip
Subject:   Re: Several gateways for two AS?

In article <16FAAE8A5.TCPMAINT@TWNMOE10.Edu.TW> TCPMAINT@TWNMOE10.Edu.TW
writes: 
     
        There are two Internet connection service providers in my country
    and we hope to establish several gateways in different locations to
    connect the two networks. Is it possible? 

Yes.

    Do we have to apply two different Aotonomous System Numbers for them? 

No.

    What routing protocol do we have to run on those gateways?

Probably BGP, but your service providers can tell you more.

Tony


-----------[000015][next][prev][last][first]----------------------------------------------------
Date:      2 May 1994 10:31:40 GMT
From:      etxmesa@eos.ericsson.se (Michael Salmon)
To:        comp.protocols.tcp-ip
Subject:   Re: Telnet option negotiation

In article <12046FIGJUTCNAITCGP@gcomm.com>
stein@gcomm.com writes:
|> 
|> I'm starting to get the feeling that there's more to Telnet option
|> negotiation that meets the eye; that it's not as straightforward as
|> pure command-reply pairs, particularly with the ECHO option.  (Are
|> there two ECHO options?  Client-echo and server-echo?  Are they
|> automatically mutually exclusive?)

Telnet is not a client-server protocol, it is intended to be as
symetrical as possible, hence the concept of client-echo and
server-echo don't apply. The NVT is presumed to echo characters typed
locally but the remote end can be requested to echo characters.
Similarly the remote end can request the local NVT to echo characters
back, I can't think of a good reason why but the point is that the
protocol allows it.

|> Does anyone know of a good description of the minimum option negotiation
|> features that Telnet clients and daemons must support?

All clients and daemons must support all options, but only to the
extent that they report that they won't do them. RFC 854 is a good
reference.

-- 

Michael Salmon

#include	<standard.disclaimer>
#include	<witty.saying>
#include	<fancy.pseudo.graphics>

Ericsson Telecom AB
Stockholm

-----------[000016][next][prev][last][first]----------------------------------------------------
Date:      Mon, 2 May 1994 10:38:42 GMT
From:      perretc@eig.unige.ch (Perret Cedric)
To:        comp.protocols.tcp-ip
Subject:   HELP PPP: etherPPP or Cameleon

Hello, 

i'm trying to use winsock based applications over PPP (with only shareware 
programs).

Actually i'm trying to do this with the etherPPP driver. (it replace the packet
driver). 

I suppose that i have to load 1)PPP, 2)winpkt, 3)Windows, 4)Mosaic.

In fact, this principe is new to me, so i've a lot of problems/questions.

1) What should i use to the other side? Have i to use a "PPP server" ?

2) if yes, such a things exist on PC? Where can i foud it?

3) I've also a SPARC-LX with solaris 2.3 connected to the network. Will i be able
   to use Mosaic (or any winsock based application) on my pc that runs PPP 
   (etherPPP driver from Merit) connected by modem to this SUN?

4) if yes, how should i install/configure PPP on solaris 2.3?


5) How can i configure etherPPP to do the login? (username and password)?


Now, i'm trying to use etherPPP, but i don't know if it is the better way to 
do this.
I've heard about Cameleon (who comes with the Windows Internet Tour Guide).
Maybe it is an easier solution to do this?
If yes, i've the same question as the number 3) with etherPPP.

6) I've also a SPARC-LX with solaris 2.3 connected to the network. Will i be able
   to use Mosaic (or any winsock based application) on my pc that runs PPP 
   (from Cameleon ) connected by modem to this SUN?


Thanks a lot to E-mail me ANY answers that can help me.

			PERRET Cedric
			PERRETC@EIG.UNIGE.CH

-----------[000017][next][prev][last][first]----------------------------------------------------
Date:      2 May 1994 17:40:24 -0400
From:      wickman@anw.cs.utk.edu (Paul J. Wickman)
To:        comp.protocols.tcp-ip
Subject:   Re: Subnet mask

In article <2q3l86$5pd@news.sprintlink.net> paul@hawksbill.sprintmrn.com writes:
>Ganesh Vaidee (ganesh@TFS.COM) wrote:
>
>> Is it possible to have a subnet mask of 255.255.0.0 
>
>Only on a "class B" address.


	Errr, It's also completely legal to have that subnet mask for a
class A address.

-- 
Paul J. Wickman - Graduate Student - University of Tennessee, Knoxville 
Internet: wickman@cs.utk.edu	

#include <disclaimer.h>

-----------[000018][next][prev][last][first]----------------------------------------------------
Date:      2 May 1994 12:55:42 GMT
From:      per@erix.ericsson.se (Per Hedeland)
To:        comp.unix.sys5.r3,comp.protocols.tcp-ip
Subject:   Re: Strange behaviour of TCP/IP in ISC 4.0

[ I'm adding comp.protocols.tcp-ip to this thread - the problem is with
  a ttcp-like program (appended at the very end of this article) sending
  from our Interactive (a.k.a. ISC) 4.0 system to various other systems,
  e.g. Sun, HP 9000, IBM RS6000 - it essentially hangs after a short
  while. Sending in the other direction works fine, as does sending
  locally on the ISC system. Furthermore, other people have reported
  being able to run the program without problems. It may well be a
  problem with our setup/configuration rather than ISC's TCP/IP, but we
  sure can't figure out what. In short, we're stumped... ]

All the possibly relevant details I can think of:
- Interactive version 4.0, with whatever version of TCP/IP that goes
  with it 
- Compaq DescPro 50M (486DX2/50)
- Ungerman-Bass 16-bit EtherNext UTP16A ethernet card, claimed to
  emulate Novell NE2000, and configured as such (ne0)
- Using either the 10BaseT (twisted-pair) or AUI (DB-15) interface on
  the card
- The Ethernet (no routing involved) is 10BaseT, with an Ungerman-Bass
  hub (and Ungerman-Bass adapters AUI<->10BaseT where needed).

One thing that I find a bit odd is that 'netstat -i' on the ISC system
claims an MTU of 1512 on the ne0 interface, whereas all other systems
use 1500 - probably not relevant though, since it appears that ISC only
ever sends 1k packets, at least with this particular program (see below).

Furthermore, netstat -i doesn't report any errors related to the running
of this program - right now, after several failed attempts, it gives
Ierrs=2 (unchanged since before these attempts), Oerrs=0, Collis=0.

Finally, we *have* been able to run the program on very rare occasions -
several of those were in the middle of the night, which would seem to
imply that the problem is load-related - however the network in question
is very lightly loaded during daytime too, rarely above 5%.

Here are parts of a tcpdump trace from running the sending side of the
program on the ISC system (raki) and the receiving side on a
SparcStation 2 with SunOS 4.1.3 (pernod). All the tcpdump traces we have
done show the same pattern. It starts out fine...

12:15:50.140971 arp who-has pernod tell raki
12:15:50.141306 arp reply pernod is-at pernod
12:15:50.141739 raki.1052 > pernod.2670: S 123136001:123136001(0) win 4096 <mss 1024>
12:15:50.142555 pernod.2670 > raki.1052: S 1676160000:1676160000(0) ack 123136002 win 4096 <mss 1460>
12:15:50.144308 raki.1052 > pernod.2670: . ack 1 win 4096
12:15:50.147671 raki.1052 > pernod.2670: P 1:1025(1024) ack 1 win 4096
12:15:50.149153 raki.1052 > pernod.2670: P 1025:2049(1024) ack 1 win 4096
12:15:50.149585 pernod.2670 > raki.1052: . ack 2049 win 4096
12:15:50.150715 raki.1052 > pernod.2670: P 2049:3073(1024) ack 1 win 4096
12:15:50.153186 raki.1052 > pernod.2670: P 3073:4097(1024) ack 1 win 4096
12:15:50.153605 pernod.2670 > raki.1052: . ack 4097 win 4096
...
12:15:50.349234 raki.1052 > pernod.2670: P 103425:104449(1024) ack 1 win 4096
12:15:50.351385 raki.1052 > pernod.2670: P 104449:105473(1024) ack 1 win 4096
12:15:50.351808 pernod.2670 > raki.1052: . ack 105473 win 4096
12:15:50.352927 raki.1052 > pernod.2670: P 105473:106497(1024) ack 1 win 4096
12:15:50.355336 raki.1052 > pernod.2670: P 106497:107521(1024) ack 1 win 4096
12:15:50.355752 pernod.2670 > raki.1052: . ack 107521 win 4096

...but then something happens; it appears that the ISC system omits a
packet:

12:15:50.357265 raki.1052 > pernod.2670: P 107521:108545(1024) ack 1 win 4096
12:15:50.359485 raki.1052 > pernod.2670: P 108545:109569(1024) ack 1 win 4096
                                                  ^^^^^^
12:15:50.367498 raki.1052 > pernod.2670: P 110593:111617(1024) ack 1 win 4096
                                           ^^^^^^
12:15:50.367739 pernod.2670 > raki.1052: . ack 109569 win 4096
12:15:50.374562 raki.1052 > pernod.2670: P 111617:112641(1024) ack 1 win 4096
12:15:50.544309 pernod.2670 > raki.1052: . ack 109569 win 4096

Right, here it comes:

12:15:51.116637 raki.1052 > pernod.2670: . 109569:110593(1024) ack 1 win 4096
12:15:51.117579 pernod.2670 > raki.1052: . ack 112641 win 3072

But after this point, throughput goes down the drain, in effect the
program grinds to a halt:

12:15:51.120835 raki.1052 > pernod.2670: . 112641:113665(1024) ack 1 win 4096
12:15:51.121265 pernod.2670 > raki.1052: . ack 113665 win 4096
12:15:51.124482 raki.1052 > pernod.2670: . 114689:115713(1024) ack 1 win 4096
12:15:51.144300 pernod.2670 > raki.1052: . ack 113665 win 4096
12:15:53.116513 raki.1052 > pernod.2670: . 113665:114689(1024) ack 1 win 4096
12:15:53.117275 pernod.2670 > raki.1052: . ack 115713 win 4096
12:15:53.120517 raki.1052 > pernod.2670: . 115713:116737(1024) ack 1 win 4096
12:15:53.144290 pernod.2670 > raki.1052: . ack 116737 win 4096
12:15:53.147498 raki.1052 > pernod.2670: . 117761:118785(1024) ack 1 win 4096
12:15:53.344290 pernod.2670 > raki.1052: . ack 116737 win 4096
12:15:57.116505 raki.1052 > pernod.2670: . 116737:117761(1024) ack 1 win 4096
12:15:57.117321 pernod.2670 > raki.1052: . ack 118785 win 4096
12:15:57.120654 raki.1052 > pernod.2670: . 118785:119809(1024) ack 1 win 4096
12:15:57.144302 pernod.2670 > raki.1052: . ack 119809 win 4096
12:15:57.147505 raki.1052 > pernod.2670: . 120833:121857(1024) ack 1 win 4096
12:15:57.344258 pernod.2670 > raki.1052: . ack 119809 win 4096
12:16:13.115946 raki.1052 > pernod.2670: . 119809:120833(1024) ack 1 win 4096
12:16:13.116818 pernod.2670 > raki.1052: . ack 121857 win 4096
12:16:13.120246 raki.1052 > pernod.2670: . 121857:122881(1024) ack 1 win 4096
12:16:13.144283 pernod.2670 > raki.1052: . ack 122881 win 4096
12:16:13.147497 raki.1052 > pernod.2670: . 123905:124929(1024) ack 1 win 4096
12:16:13.344235 pernod.2670 > raki.1052: . ack 122881 win 4096
12:16:45.114724 raki.1052 > pernod.2670: . 122881:123905(1024) ack 1 win 4096
12:16:45.144292 pernod.2670 > raki.1052: . ack 124929 win 3072
12:16:45.151444 raki.1052 > pernod.2670: P 124929:125953(1024) ack 1 win 4096
12:16:45.151991 pernod.2670 > raki.1052: . ack 125953 win 4096
12:16:45.155372 raki.1052 > pernod.2670: . 126977:128001(1024) ack 1 win 4096
12:16:45.344183 pernod.2670 > raki.1052: . ack 125953 win 4096
12:17:03.217121 pernod.2670 > raki.1052: F 1:1(0) ack 125953 win 4096
12:17:03.219141 raki.1052 > pernod.2670: . ack 2 win 4096
12:17:49.132250 raki.1052 > pernod.2670: . 125953:126977(1024) ack 2 win 4096
12:17:49.132719 pernod.2670 > raki.1052: R 1676160002:1676160002(0) win 0
12:17:49.135482 raki.1052 > pernod.2670: R 126977:126977(0) ack 2 win 4096 <mss 1024>

(I aborted it manually at this point, 100 bytes/sec is no fun...)
Needless to say, *any* help would be much appreciated!

--Per Hedeland
per@erix.ericsson.se  or
per%erix.ericsson.se@sunic.sunet.se  or
...uunet!erix.ericsson.se!per

Test program------------------------------------------------------------
#include <stdio.h>
#include <sys/types.h>
#include <sys/socket.h>
#include <netinet/in.h>
#include <sys/time.h>
#include <netdb.h>

struct sockaddr_in iserv_addr;

/* Call as "a.out -s" on a non-ISC system
 *         "a.out hostname portnum" on the ISC system
 */
main(argc, argv)
int argc;
char **argv;

{
    if (strcmp(argv[1], "-s") == 0)
	server();
    else 
	client(argv[1],argv[2]);
}
 
server()
{
    int length, rval,msgsock,s;
    char buf[1024];
    
    if ((s  = socket(AF_INET,SOCK_STREAM,0)) < 0) err();
    iserv_addr.sin_family = AF_INET;
    iserv_addr.sin_addr.s_addr = htonl(INADDR_ANY);
    iserv_addr.sin_port = 0;
    if (bind(s ,(struct sockaddr*) &iserv_addr,sizeof(iserv_addr)) <0) 
	err();
    listen(s ,5);
    
    /* find out assigned portnumber */
    length = sizeof(iserv_addr);
    if (getsockname(s, (struct sockaddr *)&iserv_addr, &length) < 0) 
	err();
    printf("Port = %d\n", ntohs(iserv_addr.sin_port));
    if ((msgsock = accept(s ,(struct sockaddr*)0,(int*)0)) < 0)
	err();
    while ( read(msgsock, buf, 1024 ) > 0) ;
    printf("Done\n");
    exit(1);
}


client(host, port)
char *host, *port;
{
    struct hostent *hp;
    int i, s, cport;
    char buf[1024];
    struct timeval tv, inittv;
    long milliseconds;


    if (sscanf(port,"%d", &cport) != 1)
	exit(1);
    if ((s = socket(AF_INET,SOCK_STREAM,0)) <= 0)
	err();
    if ((hp = gethostbyname(host)) == NULL)
	err();
    memcpy((char*)&iserv_addr.sin_addr,
	   (char*)hp->h_addr, hp->h_length);
    iserv_addr.sin_family = AF_INET;
    iserv_addr.sin_port = htons(cport);

    if(connect(s , (struct sockaddr*)&iserv_addr, sizeof iserv_addr) <0) 
	err();
    gettimeofday(&inittv,(struct timezone*)0);
    for (i=0; i< 1000; i++) {
	if (write(s, buf, 1024) != 1024)
	    err();
    }
    gettimeofday(&tv,(struct timezone*)0);
    milliseconds = 1000 * (tv.tv_sec - inittv.tv_sec) +
        (tv.tv_usec - inittv.tv_usec) / 1000;
    printf("1 Meg in %d msecs \n", milliseconds);
}

err()
{
    perror("");
    exit(1);
}


-----------[000019][next][prev][last][first]----------------------------------------------------
Date:      Mon, 2 May 1994 12:57:46 GMT
From:      abstine@hummer.visix.com (Art Stine)
To:        comp.protocols.tcp-ip
Subject:   Re: Telnet option negotiation

In article <12046FIGJUTCNAITCGP@gcomm.com> stein@gcomm.com writes:


   I'm starting to get the feeling that there's more to Telnet option
   negotiation that meets the eye; that it's not as straightforward as
   pure command-reply pairs, particularly with the ECHO option.  (Are
   there two ECHO options?  Client-echo and server-echo?  Are they
   automatically mutually exclusive?)

   Does anyone know of a good description of the minimum option negotiation
   features that Telnet clients and daemons must support?

How about checking the RFCs ? for sample implementations, you can probably look
at the BSD code, CUTCP code, etc to see what other folks have done (though
that won't give you a 'minimum' set but more likely the common/full set).

--
-art

----                           __
Art Stine               \  / /(_  /\/  11440 Commerce Park Drive
Software Engineer        \/ / __)/ /\  Reston, Virginia 22091
abstine@visix.com       Software Inc   Voice: (703) 758-2767 (direct)
                                              (703) 758-8230 (main)
"Spambi, Spambi, as much fun as               (800) 832-8668 (toll-free)
potted meat can-be..."                 FAX:   (703) 758-0233


-----------[000020][next][prev][last][first]----------------------------------------------------
Date:      2 May 1994 14:36:09 GMT
From:      hill@raleng.mtc.com (Mike Hill)
To:        comp.protocols.tcp-ip
Subject:   Telnet Testing

Hi,

    I'm looking for a some sort of Telnet testing application
which can open multiple telnet client sessions.  I know
that there are Telnet apps. out ther that can open 10 or so sessions,
but I need 100's of sessions for testing.  If you are aware of
something that can do this please let me know.

Thanks.
MJH

-----------[000021][next][prev][last][first]----------------------------------------------------
Date:      Mon, 02 May 94 16:32:37 GMT
From:      TCPMAINT@TWNMOE10.Edu.TW
To:        comp.protocols.tcp-ip
Subject:   Several gateways for two AS?

Hi!
 
    There are two Internet connection service providers in my country
and we hope to establish several gateways in different locations to
connect the two networks. Is it possible? Do we have to apply two
different Aotonomous System Numbers for them? What routing protocol
do we have to run on those gateways?
 
Thanx,
 
/Liu Zi-Di

-----------[000022][next][prev][last][first]----------------------------------------------------
Date:      2 May 1994 23:49 MST
From:      kaplan@bpavms.bpa.arizona.edu (RayK)
To:        comp.protocols.tcp-ip
Subject:   Risk reducer

Cross post to RISKS (via mail submission), comp.security.announce and
comp.protocols.tcp-ip news groups, and a few other various places.  Sorry
if you see this more than once.

Re: Firewalls and Internet Security - Repelling the Wily Hacker.
Ray Kaplan - May 2, 1994

Buy this book

Gentle folk,

Here is a risk reducer.  With the wholesale rush to Internet connectivity, its
about time someone sat down and wrote a good book about how to do this
exercise safely!  And, sure enough, Cheswick and Bellovin have done just that,

Heaping superlatives on something of which you are enamored is always
problematic - the possibility of overstatement looms large.  Accordingly I`ll
cut to the chase.  Buy this book!  I do not get any money for saying this - I
just believe you are well justified in getting it on your reading list -
today. In May of this year, Addison Wesley is releasing an excellent new book
by Bill Cheswick and Steve Bellovin: Firewalls and Internet Security -
Repelling the Wily Hacker.  ISBN 0-201-63357-4.  It will retail for $26.95. 
Bulk purchases: 800- 238-9682, individual orders: 800-824-7799 (FAX
617-944-7273).  Email orders over the Internet from bexpress@aw.com (no they
don`t take plastic via Email). For those that are net-challenged, U.S.
snailmail orders from Addison-Wesley, c/o Arlene Morgan, 1 Jacob Way, Reading,
MA  01867 USA. 

Rumors loom large that at least one of the authors (Ches?) will be at Interop
with copious quantities of this work of art.  As dues of superlative
authorship that is destined to be popular, I hope they both get writer`s cramp
autographing! 

Details 

While worthwhile, well written, pace-setting, technically astute works of art
are rare - this is certainly one of them.  I am always hard pressed to
identify any one thing as unique in its decade (especially when the decade is
still in progress). Suffice it to say that this work is the most complete
treatment of firewall technology and experience that is available.  The
availability of this work is exciting news for security firewall builders -
including Internet security firewall builders - and, for the great number of
people that seem to be befuddled by the complexity and the general issues of
interconnecting networks. 

The book 

While my review copy (well dog-eared, now) is a bit dated (March 7, 1994), I
think you can expect that it is close to the book`s final form: a standard
(w=7.5in, h=9in) Addison-Wesley Professional Computing Series book like the
ones that should already dot your shelves.  (I don`t get any money for my
obvious favorable bias toward this series.  My bias is born out of the fact
that the series (Brian Kernighan is the consulting editor for it) contains
great authors and titles like Radia Pealman`s Interconnections - Bridges and
Routers and Richard Sevens` TCP/IP Illustrated, Volume I - The Protocols.) 

305 pages in 14 chapters, appendices, a bibliography, a list of "bombs"
(security holes) and an index. 

Out of the box, the authors set the tone for their work by quoting F.T. Gramp
and R.H. Morris: "It is easy to run a secure computer system.  You merely
have to disconnect all dial-up connections and permit only direct-wired
terminals, put the machine and the terminals in a shielded room, and post a
guard at the door."  This is followed by a detailed discussion of the art
and science of building a firewall. There is so much good stuff here, that all
I can do is list the book`s contents - lest I write a tome which distracts you
from picking up a copy of it ASAP. 

Chapters and content - from the table of contents.

Getting started
Introduction
- Why security?
- Picking a security policy 
- Strategies for a secure network
- The ethics of computer security
- Warning
Overview of TCP/IP
- The different layers
- Routers and routing protocols
- The Domain name service
- Standard services
- RPC-based protocols
- The "r" commands
- Information services
- The X-11 service
- Patterns of trust

Building your own firewall
Firewalls and gateways
- Firewall philosophy
- Situating firewalls
- Packet-filtering gateways
- Application-level gateways
- Circuit-level gateways
- Supporting inbound services
- Tunnels - good and bad
- Joint Ventures
- What firewalls can`t do
How to build an application-level gateway
- Policy
- Hardware configuration options
- Initial installation
- Gateway tools
- Installing services
- Protecting the protectors
- Gateway administration
- Safety analysis - why our setup is secure and fail-safe
- Performance
- The TIS firewall toolkit
- Evaluating firewalls
- Living without a firewall
Authentication
- User authentication
- Host-to-host authentication
Gateway tools
- Proxylib
- Syslog
- Watching the network: Tcpdump and friends
- Adding logging to standard demons
Traps, lures and honey pots
- What to log
- Dummy accounts
- Tracing the connection
The hacker`s workbench
- Introduction
- Discovery
- Probing hosts
- Connection tools
- Routing games
- network monitors
- Metastasis
- Tiger teams
- Further reading

A look back
Classes of attacks
- Stealing passwords
- Social engineering
- Bugs and backdoors
- Authorization failures
- Protocol failures
- Information leakage
- Denial-of-service
An evening with Berferd
- Introduction
- Unfriendly acts
- An evening with Berferd
- The day after
- The jail
- Tracing Berferd
- Berferd comes home
Where the wild things are: a look at the logs
- A year of hacking
Proxy use
- Attack sources
- Noise on the line

Odds and ends
Legal considerations
- Computer crime statutes
- Log files as evidence
- Is monitoring legal?
- Tort liability considerations
Secure communications over insecure networks
- An introduction to cryptography
- The Kerberos authentication system
- Link-level encryption
- Network- and transport-level encryption
- Application-level encryption

Where do we go from here? 
Appendices

Useful free stuff 
- Building firewalls
- Network management and monitoring tools
- Auditing packages
- Cryptographic software
- Information sources

TCP and UDP ports - Fixed ports
- MBone usage

Recommendations to vendors
- Everyone
- Hosts
- Routers
- Protocols
- Firewalls

Bibliography - List of bombs - Index 

I have criticisms, complaints and suggestions.  However, considering that this
is such a darn fine piece of work - I hasten to get my recommendation that you
buy this book out ASAP. 

Meantime, to whet your appitite: 

- Index - (a well done, 26 pages worth - you can actually find pointers to
what you want to know!  What a concept. 

- TCP ports discussion - a Comprehensive list and reasonable advice on what to
do with them. 

- Bombs - a summarized list of the 43 major security holes that they identify.

- Bibliography - Ahhhh.  19 pages of the best firewalls-related bibliography
that I`ve seen. 

- Where to from here - excellent advice for techies and managers who don`t
want to keep working at the job of firewalling or who simply want to spend a
bit of resources on it only once. 

Kudos to the authors - buy this book. 

Of course - these are my own views, and they don`t necessarily reflect those
of anyone - including my employer.  However, in this case, they probably do. 

----------
Ray Kaplan             CyberSAFE, Corporation
rayk@ocsg.com          Formerly Open Computing Securyt Group (OCSG)
                       (206) 883-8721
                       FAX at (206) 883-6951
                       2443 152nd Ave NE
                       Redmond, WA 98052

Better living through authentication
---------

-----------[000023][next][prev][last][first]----------------------------------------------------
Date:      2 May 1994 17:23:14 GMT
From:      mcochran@wellfleet.com (Marc Cochran)
To:        comp.protocols.tcp-ip
Subject:   Re: Destination Unreachable

In article 8il@cmcl2.NYU.EDU, roy@mchip00.med.nyu.edu (Roy Smith) writes:
	other stuff deleted
> 
> 	When I ping from 205.115 to 205.14, I get Desination Unreachables
> and I don't understand why (see ping output below).  One came from 127.0.0.1
> (i.e.  the loopback interface), and others from the two SGI's (205.96 and
> 135.9).  What does this mean?  What I expected was to just not get any
> response from the powered-off host, and can't figure out what the Dest
> Unreachables mean and why I'm getting them.
> 
> mchip00> ping -v 128.122.205.14
> PING 128.122.205.14: 56 data bytes
> 36 bytes from 127.0.0.1: icmp_type=3 (Dest Unreachable)

Some routers generate Destination Unreachables not just when there is no route
available, but when they don't receive an ARP REPLY in response to an ARP REQUEST
packet.  I know of at least 1 router which does this.  The router would send 5
ARP REQUESTs 1 second apart, and if it didn't receive a REPLY, it would then
generate the unreachable.  Do you see the unreachable immediately, or does it
take a few seconds ?

Marc

-----------[000024][next][prev][last][first]----------------------------------------------------
Date:      2 May 1994 18:25:37 GMT
From:      henryh@well.sf.ca.us (Henry Hwong)
To:        comp.protocols.tcp-ip
Subject:   Nonblocking TCP sockets

------------------------------------------------------------------------
I am forwarding these questions to the net for someone w/o news access.
Please answer via mail to Peter_Schonberger@notes.bsginc.com

I can forward a summary if there are enough requests...

-Henry (henryh@well.sf.ca.us / Henry_Hwong@notes.bsginc.com)
------------------------------------------------------------------------

I have two questions pertaining to using TCP sockets in non-blocking mode.
The question is in the specific context of HP-UX, namely HP 9000 S800
w/HP-UX 9.04.

Briefly, the first question is "how do you set your socket non-blocking?",
while the second one is "how do you use your non-blocking socket to
connect?".

QUESTION #1
===========

The 'man 7 socket' helps:

  "The fcntl(2) O_NDELAY and ... flags ... are supported by sockets.
   ...  If the O_NDELAY flag is set, the socket is put into non-blocking
  mode."

One can set it:

  fcntl( sfd, F_SETFL, O_NDELAY);

and read it with F_GETFL, except I'm not sure about the third arg in the
calling sequence.  I found that this works:

  fflags = fcntl( sfd, F_GETFL, 0 );

  if( (fflags & O_NDELAY) == O_NDELAY ) ...

Is this the way it is supposed to be ???

Confusion starts when fcntl() and ioctl() are mixed.  'man 5 ioctl' tells
me that

  "FIOSNBIO ... does not interfere with AT&T O_NDELAY open and fcntl flag"

'man 2 open' only tells me about O_NDELAY that on communication lines we
will block/not block waiting for carrier, whether O_NDELAY is off/on.
So, I presume that for sockets fcntl() must be setting something in the
socket object, not the generic file.  To use ioctl(), 'man 5 ioctl' spells
out clearly what I have to do:

  int dummy = 1;

  rc = ioctl( sfd, FIOSNBIO, &dummy );

This executes without complaints, but I cannot check directly what happened
since HP's FIOGNBIO apparently does not work for sockets.  The call

  rc = ioctl( sfd, FIOGNBIO, &dummy );

returns -1, with strerror( errno ) being:

  "Can't assign requested address".

So, I suppose that O_NDELAY and FIOSNBIO are different things.  What are
they?  Man page 'man 5 ioctl' defines clearly the read/write behaviour under
the control of FIOSNBIO.  Then the last clarification given is the following:

  To prohibit non-blocking I/O from interfering with the O_NDELAY flag (see
  open(2) and fcntl(2)", the functionality of O_NDELAY always supercedes the
  functionality of non-blocking I/O.  This means that if O_NDELAY is set, the
  driver performs read requests in accordance with the definition of O_NDELAY.
  When O_NDELAY is not set, the definition of non-blocking I/O applies.

All this seems to say that AT&T inherited O_NDELAY is supported for sockets
to put them in "non-blocking mode".  But, HP's definition of "non-blocking
I/O" is supported by FIOSNBIO, which must be used with O_NDELAY clear.

To summarize, the question is:

What's the difference between "non-blocking mode" (that is, O_NDELAY) vs.
"non-blocking I/O" (that is, FIOSNBIO)?  I know the statement that usually,
fcntl() operates at file level, while ioctl() may access the device, but I do
bot grasp the implication for the socket code.  Is ioctl(FIOSNBIO, ..) the
"right" way?  Am I doing something wrong about FIOGNBIO?  If my socket code
is to compile and run on an AT&T host as well, is using fcntl(O_NDELAY,...)
just as good and portable as well?

QUESTION #2
===========

'man 2 connect' is not of much help either to choose between O_NDELAY and 
FIOSNBIO, but it indicates how to connect:

  The call normally blocks until the connection completes.  If non-blocking
  mode has been enabled using ... O_NDELAY fcntl() flags or the FIOSNBIO
  ioctl() request and the connection cannot be completed immediately,
  connect() returns an error

The errno returned is EINPROGRESS and

  ... this is not a failure.  Make the connect() call again a few seconds
  later.  Alternatively, wait for completion by calling seelct(), selecting
  for write.

That is, when all goes right connect( sfd, ... ) returns -1 with errno set
to EINPROGRESS and then select() with the write-mask including sfd will block
until the connection is established.  But, what happens when things go wrong?

So, the second question is:

How is the failure to connect detected (including the various possible errno
values)? When the local host knows that things are wrong, things are simple.
Namely, I expect connect() to return immediately -1 with errno set to
EALREADY, EISCONN, ...  But how do I collect error cases reported later,
after the three-way handshake, by the remote host, like ETIMEDOUT,
ECONNREFUSED, ...  Or, network-down, host-down, ... situations?  Should I
include the connecting sfd in the read-mask of select as well?  If so, 
will I ever get these errno-s or what?

-Peter Schonberger (Peter_Schonberger@notes.bsginc.com)

-----------[000025][next][prev][last][first]----------------------------------------------------
Date:      2 May 1994 18:45:16 GMT
From:      scotta@kije.gsfc.nasa.gov (Scott Austin)
To:        comp.protocols.tcp-ip,misc.forsale.computers.other
Subject:   Re: Bargains on Comer's 3 Volume TCP/IP set

Jeff Macdonald <jeff@zis.ziff.com> wrote:

>However the three books could not total over $100 or $105.

Normally to join LCIS, you pick (at best) 4 books for $1 each with the total
publisher's price not exceeding $110.  AND with that offer, you have to buy *3*
selections over the next 2 years.  I've also seen a [deep breath] 3-books-for-
$1-each/total-less-than-$97.50/gotta-buy-3-books-over-the-next-year [exhale] 
offer.

Compare that to this offer for Comer's 3 volume set (pub. price: $164, for
$4.95 with the obligation of buying only 1 book in the next year).

Scott Austin
scotta@cnt.com

-----------[000026][next][prev][last][first]----------------------------------------------------
Date:      2 May 1994 19:45:10 GMT
From:      paul@sprintlink.net (Paul Ferguson)
To:        comp.protocols.tcp-ip
Subject:   Re: Subnet mask

Ganesh Vaidee (ganesh@TFS.COM) wrote:

> Is it possible to have a subnet mask of 255.255.0.0 

Only on a "class B" address.

_______________________________________________________________________________
Paul Ferguson                         
US Sprint 
Enterprise Internet Engineering                    tel: 703.904.2437 
Herndon, Virginia  USA                        internet: paul@hawk.sprintmrn.com

-----------[000027][next][prev][last][first]----------------------------------------------------
Date:      Mon, 2 May 1994 20:41:42 GMT
From:      donp@novell.com (don provan)
To:        comp.protocols.tcp-ip
Subject:   Re: What constitutes a duplicate TCP packet?

In article <cearley_k.33.000EC06D@wizard.colorado.edu> cearley_k@wizard.colorado.edu (Kent Cearley) writes:
>We have a problem with a host sending two tcp acknowledgements... one with 
>WIND=0 and no data, followed immediately with a WIND=256, also no data. Both 
>the sequence numbers and ack numbers are the same on both packets. The problem 
>is, a client (FTP Inc.'s WTN3270) sees the first packet, assumes it cannot 
>transmit, and appears to be discarding the second packet without the WIND=256 
>registering.
>
>FTP Inc. claims the RFC specifies if the SEQ# are the same its a duplicate 
>packet. If so, how is this situation handled? All of our other clients 
>recognize the second packet as advertising an available window. 

I'm sorry, but I don't believe you. I just can't believe FTP, Inc.,
would say that. I mean, after all, right now you're holding in your
hands two packets that prove unequivocally that packets having the
same sequence number are not always duplicates. It's true that the TCP
spec itself has some fuzzy writing -- or fuzzy thinking, depending on
who you ask -- in describing the handling of window, ACK, and urgent
information in packets containing stale sequence numbers, but cases
such as yours -- which, by the way, is quite common -- have made it
obvious since way back that its important to process this information
even when it is received in packets that don't have fresh sequence
numbers.

						don provan
						donp@novell.com

-----------[000028][next][prev][last][first]----------------------------------------------------
Date:      2 May 1994 20:52:24 GMT
From:      slama@ctron.com (Frederick G. Slama)
To:        comp.protocols.tcp-ip
Subject:   OSF DCE licensing


	How?

	I would like to acquire the DCE developement tools for SunOS4.1.3.
	Is this possible? And at what cost?

	Thanks,
	-Fred

-----------[000029][next][prev][last][first]----------------------------------------------------
Date:      Mon, 2 May 1994 20:53:51 GMT
From:      donp@novell.com (don provan)
To:        comp.protocols.tcp-ip,comp.protocols.tcp-ip.ibmpc
Subject:   Re: Problem with FTPing to RS6000 with Novell LWP for DOS...HELP!!!

In article <CoznsI.p0r@sernews.raleigh.ibm.com> mebane@ralvm12.vnet.ibm.com writes:
>I'll take a stab since I have worked on our ftp and ftpd server quite a
>bit. The 425 imples that the server cannot open the data connection to
>the client. Lots of possible reasons, but often because of an improper
>socket setup.

I missed the initial question, so I'm really in the dark about this
one, but the times I've see trouble opening the data connection, the
cause has been a deficient FTP server that ignored the PORT command. I
don't remember exactly which server implementation it was, but I seem
to recall it was, in its day, a fairly common UNIX implementation.

						don provan
						donp@novell.com

-----------[000030][next][prev][last][first]----------------------------------------------------
Date:      2 May 1994 21:35:46 GMT
From:      barmar@think.com (Barry Margolin)
To:        comp.protocols.tcp-ip
Subject:   Re: Subnet mask

In article <Cp1ML9.EKE@tfs.com> ganesh@TFS.COM (Ganesh Vaidee) writes:
>Is it possible to have a subnet mask of 255.255.0.0 

Sure.

>Does this mean that there are o bits for subnet id and 0 bits for host id.
>How is that possible. help. thanx

No, it means that there are 16 bits of network and subnet, and 16 bits of
host.  The boundary between network and subnet depends on whether the
network is class A or B.  And for systems that support supernetting, this
mask could be used for a block of 256 class C networks.
-- 
Barry Margolin
System Manager, Thinking Machines Corp.

barmar@think.com          {uunet,harvard}!think!barmar

-----------[000031][next][prev][last][first]----------------------------------------------------
Date:      Mon, 2 May 1994 22:16:25 GMT
From:      jas@talking.COM (Jim Shankland)
To:        comp.protocols.tcp-ip
Subject:   Re: easy way to quickly detect a node has rebooted?

In article <Cp3zws.39q@talarian.com> scott@talarian.com (Scott Weitzenkamp) writes:
>If I use my own application-level "keepalives", which send()s data
>occasionally and then recv()s some data back from the other process, I
>can detect the reboot fairly quickly.  This works OK in most of my
>situations, but I have a few cases where the other process will be off
>in a potentially time-consuming loop.

If the other machine has rebooted, your send() will cause the
other machine to generate a RST packet, which will turn into
an ECONNRESET error for your recv().  You only need a timeout
on your recv() if you also want to be able to detect that the
peer process is still up, but "hung".  I'd say application-level
keepalives are the way you should go.

jas

-----------[000032][next][prev][last][first]----------------------------------------------------
Date:      Mon, 2 May 1994 23:28:31 GMT
From:      abell@velveeta.apdev.cs.mci.com (Andrew_Bell)
To:        comp.protocols.tcp-ip
Subject:   Handling adapter failure

I have a machine homed to two subnets.  Each subnet has a router that can
get anywhere.  What I would like is to provide some measure of
fault-tolerance without running a routing protocol between my box and the
routers.  What I would like is to set up a "preferred" default route out
one adapter to one of the routers, and a "standby" default route to the
other router on the other subnet should the first adapter go belly-up
(receives a "network unreachable", or something like that).  Is there an
easy way to do this?  Is there a hard way to do this?  Should I just give
in and run a real routing protocol?

Andrew Bell
MCI Communications
abell@chong.nms.cs.mci.com

-----------[000033][next][prev][last][first]----------------------------------------------------
Date:      2 May 1994 23:58:58 GMT
From:      paul@sprintlink.net (Paul Ferguson)
To:        comp.protocols.tcp-ip
Subject:   Re: Subnet mask

Paul J. Wickman (wickman@anw.cs.utk.edu) wrote:

> >
> >> Is it possible to have a subnet mask of 255.255.0.0 
> >
> >Only on a "class B" address.
 
> 	Errr, It's also completely legal to have that subnet mask for a
> class A address.

Right you are.

It's just a very *rare* occurance to work within "class A" address
space.  ,-)


_______________________________________________________________________________
Paul Ferguson                         
US Sprint 
Enterprise Internet Engineering                    tel: 703.904.2437 
Herndon, Virginia  USA                        internet: paul@hawk.sprintmrn.com

-----------[000034][next][prev][last][first]----------------------------------------------------
Date:      Tue, 3 May 1994 01:36:32 GMT
From:      schaumann@oms2.usmc.mil (David K Schaumann)
To:        comp.protocols.tcp-ip
Subject:   Re: Subnet mask

>> Is it possible to have a subnet mask of 255.255.0.0 
>
Paul Ferguson wrote:

>Only on a "class B" address.

Sorry, but I have to correct you.  That is, only on a "class B and a class A".  

But I find it hard to believe that anyone needs that many nodes on a class
B or a class A.  I have 3 class C's and 1 class B, and I have an 8 bit subnet 
on my class B and a 6 bit subnet on my class C's.

Dave

David Schaumann
Okinawa   Japan
InterNet: schaumann@oms2.usmc.mil
or: schaumann@okr-smtp1.usmc.mil 

-----------[000035][next][prev][last][first]----------------------------------------------------
Date:      Tue, 3 May 1994 02:16:24 GMT
From:      herbr@netcom.com (Herb Rosenberg)
To:        comp.protocols.tcp-ip
Subject:   IP Addresses and the Internet?


I have a wide area network with about 800 PC's accross 40 locations.  We 
are moving to TCP/IP and all of our routers up to this point have had an 
arbitrary IP assignment.  We are collecting information on getting a 
coordinated IP address from the INTERNIC  However, there is a debate 
amongst our technical group as to whether a "coordinated IP address block 
is needed, beyond 1 class C address for the HOST gateway that will have 
access to the Internet itself, beyond a firewall.

By questions are:

Does this makes sense, and / or all there any other issues that need to 
be considered?

If the rest is our network is "private" is there any need to get a 
coordibated address for it?

What aboout all of these large organizations that have Class B 
assignments?  Surely they have a firewall to prevent authorized access 
into there systems, so why would they need a coordinated IP address?

Is there any RFC's that talk about firewalls specifically?

Thanks.

-- 
herbr@netcom.com

-----------[000036][next][prev][last][first]----------------------------------------------------
Date:      3 May 1994 02:41:17 GMT
From:      scip3144@leonis.nus.sg (Na)
To:        comp.protocols.tcp-ip
Subject:   HELP - ST II software


Are there any FTP sites having ST II software configurable over ATM.

Thanks
-gurudev

vgurudev@iscs.nus.sg

-----------[000037][next][prev][last][first]----------------------------------------------------
Date:      3 May 1994 10:00:26 -0400
From:      purcell@zeus.IntNet.net (Kevin Purcell)
To:        comp.protocols.tcp-ip
Subject:   Resetting FTP TCP window size


I am currently running Interactive 386 Sys V R 3.2 and I need to reset the
TCP window size to a smaller value.  I need to know if I could do this through
kernel configuration parameters (reset the default TCP window size) or do I
need to modify the FTP source code to do a setsockopt system call.  The only
problem is I do not have source code for Sys V FTP.  If anyone has source
code they have written or know of an easier way.  Please respond or E-Mail me.

--



------------------------------------------------------------------------------
| God is dead - Nietzche | 		Purcell@zeus.intnet.net
| Nietzche is dead - God | Don't settle for shampoo, Demand the REAL poo..
------------------------------------------------------------------------------

-----------[000038][next][prev][last][first]----------------------------------------------------
Date:      Tue, 3 May 1994 10:05:00
From:      cearley_k@wizard.colorado.edu (Kent Cearley)
To:        comp.protocols.tcp-ip
Subject:   Re: What constitutes a duplicate TCP packet?

In article <1994May2.204142.23311@novell.com> donp@novell.com (don provan) writes:
>I'm sorry, but I don't believe you. I just can't believe FTP, Inc.,
>would say that. I mean, after all, right now you're holding in your
>hands two packets that prove unequivocally that packets having the
>same sequence number are not always duplicates. It's true that the TCP
>spec itself has some fuzzy writing -- or fuzzy thinking, depending on
>who you ask -- in describing the handling of window, ACK, and urgent
>information in packets containing stale sequence numbers, but cases
>such as yours -- which, by the way, is quite common -- have made it
>obvious since way back that its important to process this information
>even when it is received in packets that don't have fresh sequence
>numbers.

Yeah, we couldn't believe it either... FTP Inc. typically provides pretty high 
levels of support... but, two tech support contacts in a row tried to dismiss 
our problem as vendor specific due to these 'duplicate packets'. Hopefully 
they will get our trace to the right people, it was a learning experience for 
us! Thanks for the info on this list  (thanks Barry).

-Kent Cearley
-University of Colorado, Boulder


-----------[000039][next][prev][last][first]----------------------------------------------------
Date:      03 May 94 13:22 PDT
From:      Chip Benowitz <chipb@igc.apc.org>
To:        comp.protocols.tcp-ip
Subject:   Re: Port #'s for WAIS,Mosaic?

Mosaic is based on the following ports/services: (along with others?)

23	telnet
21	ftp
70	gopher
80	http

-----------[000040][next][prev][last][first]----------------------------------------------------
Date:      Tue, 3 May 1994 06:34:06 GMT
From:      jgong@netcom.com (John Gong)
To:        comp.protocols.tcp-ip
Subject:   Port #'s for WAIS,Mosaic?

Greetings,

Can anyone here in netland tell me the port nos. for these services?  I'm
trying to help a client set up router access lists based on TCP ports and
could not find them in the well known addresses RFC.

Please email replies to me at   jgong@netcom.com

Much appreciated!

-----------[000041][next][prev][last][first]----------------------------------------------------
Date:      Tue, 3 May 1994 12:30:40 EST
From:      Ray Hunter <RHUNTER@ESOC.BITNET>
To:        comp.protocols.tcp-ip
Subject:   Re: Several gateways for two AS?

You are looking to become a multi-homed AS by the sound of it....
We are multi-homed to NASA PSCNI, NSN, NL-NET, DE-WIN, ES-REDIRIS &
IT-GARR, which makes for a great deal of fun ;-)

Are you really *sure* you want to become multi-homed, as it adds a
huge amount of complexity to your routing decisions. If you can
get away with 2 redundant links to one provider I'd recommend this
route. (awful pun intended)

If you're determined to go multi-homed then you need to:

1. obtain an AS number of your own
2. start reading at least about BGP-3 (RFC 165 1266 & 1267)
3. start reading about BGP-4 if your router supports it (was an internet
   draft last time I looked.....)
   Cisco only supports it in Beta release 10... not before
4. Think long & hard about how big your routers memory & processor are
   Cisco requires a CSC/4 for BGP
5. Think long and hard about a Routing Policy...
   what routes do you want to accept
   what routes do you want to reject
   do you want to prefer internal routes over external
   do you want to redistribute BGP into your IGP
   what do you want to do about fail-over/metrics
   do you want to become a transit network
6. discuss & agree all of the above with the Internet Providers.
7. write your configs & implement

As you can see the way is long and not very well lit.

If you want to discuss anything off-line then drop me a note.

Ray Hunter
rhunter@esoc.bitnet

-----------[000042][next][prev][last][first]----------------------------------------------------
Date:      3 May 1994 07:43:43 GMT
From:      tli@cisco.com (Tony Li)
To:        comp.protocols.tcp-ip
Subject:   Re: Handling adapter failure

In article <Cp77vJ.6Lz@sgi1.fnet.cs.mci.com>
abell@velveeta.apdev.cs.mci.com (Andrew_Bell) writes: 
    I have a machine homed to two subnets.  Each subnet has a router that can
    get anywhere.  What I would like is to provide some measure of
    fault-tolerance without running a routing protocol between my box and the
    routers.  What I would like is to set up a "preferred" default route out
    one adapter to one of the routers, and a "standby" default route to the
    other router on the other subnet should the first adapter go belly-up
    (receives a "network unreachable", or something like that).  Is there an
    easy way to do this?  Is there a hard way to do this?  Should I just give
    in and run a real routing protocol?
    
You should be able to do this with Router Discovery.

Tony

-----------[000043][next][prev][last][first]----------------------------------------------------
Date:      3 May 1994 07:45:24 GMT
From:      tli@cisco.com (Tony Li)
To:        comp.protocols.tcp-ip
Subject:   Re: IP Addresses and the Internet?

In article <herbrCp7FnC.JME@netcom.com> herbr@netcom.com (Herb Rosenberg)
writes: 
    
    I have a wide area network with about 800 PC's accross 40 locations.  We 
    are moving to TCP/IP and all of our routers up to this point have had an 
    arbitrary IP assignment.  We are collecting information on getting a 
    coordinated IP address from the INTERNIC  However, there is a debate 
    amongst our technical group as to whether a "coordinated IP address block 
    is needed, beyond 1 class C address for the HOST gateway that will have 
    access to the Internet itself, beyond a firewall.
    
    By questions are:
    
    Does this makes sense, and / or all there any other issues that need to 
    be considered?
    
    If the rest is our network is "private" is there any need to get a 
    coordibated address for it?
    
    What aboout all of these large organizations that have Class B 
    assignments?  Surely they have a firewall to prevent authorized access 
    into there systems, so why would they need a coordinated IP address?
    
    Is there any RFC's that talk about firewalls specifically?
    
I suggest you look at RFC 1597 as it deals with most of these questions.

Tony



-----------[000044][next][prev][last][first]----------------------------------------------------
Date:      3 May 1994 10:55:58 GMT
From:      etxmesa@eos.ericsson.se (Michael Salmon)
To:        comp.protocols.tcp-ip
Subject:   Re: IP Addresses and the Internet?

In article <herbrCp7FnC.JME@netcom.com>
herbr@netcom.com (Herb Rosenberg) writes:
|> 
|> I have a wide area network with about 800 PC's accross 40 locations.  We 
|> are moving to TCP/IP and all of our routers up to this point have had an 
|> arbitrary IP assignment.  We are collecting information on getting a 
|> coordinated IP address from the INTERNIC  However, there is a debate 
|> amongst our technical group as to whether a "coordinated IP address block 
|> is needed, beyond 1 class C address for the HOST gateway that will have 
|> access to the Internet itself, beyond a firewall.
|> 
|> By questions are:
|> 
|> Does this makes sense, and / or all there any other issues that need to 
|> be considered?
|> 
|> If the rest is our network is "private" is there any need to get a 
|> coordibated address for it?

Every machine needs to be able to unambiguously address every other
machine that it wishes to communicate with, I think that that is
axiomatic. Hence if your gateway is the only machine that communicates
with the real world then you only need one address but that also means
that all traffic must terminate there and then be forwarded to the user.

|> What aboout all of these large organizations that have Class B 
|> assignments?  Surely they have a firewall to prevent authorized access 
|> into there systems, so why would they need a coordinated IP address?

I live behind a firewall and Ericsson's is sufficiently large that it
has several class B and C addresses. I guess that we could have built a
system as you have described but I don't think that it would be
pleasant to use and a nightmare to maintain. I am currently trying to
get access to the real world for WWW, with our setup I only have our
security group to fight, I don't need to fight the network as well.

|> Is there any RFC's that talk about firewalls specifically?

None that I know of, the closest I could quickly find was 1579.

-- 

Michael Salmon

#include	<standard.disclaimer>
#include	<witty.saying>
#include	<fancy.pseudo.graphics>

Ericsson Telecom AB
Stockholm

-----------[000045][next][prev][last][first]----------------------------------------------------
Date:      Tue, 03 May 1994 21:44:00 -0500
From:      Bill.Beers@f240.n379.z1.fidonet.org (Bill Beers)
To:        comp.protocols.tcp-ip
Subject:   Re: IP RIP Question in multi-subnet/network environment

 AF> From: albert@dn.itg.telecom.com.au (Albert Fu) 
 AF> Organization: Telecom Australia 
 AF> 
 AF> (fictitious network used, netmask 255.255.255.0) 
 AF> 
 AF>   +-----------+                            +-----------+ 
 AF>   |  Router1  |                            |  Router2  | 
 AF>   +-----------+                            +-----------+ 
 AF>         |150.150.1.1                             |150.150.1.2 
 AF>         |150.150.2.1                             | 
 AF>         |                                        | 
 AF>     |------------------------------------------------| 
 AF> 
 AF> Both routers run the RIP protocols, and have connections to other 
 AF> networks/subnets. During the periodic RIP update (every 30s), Router1 
 AF> sends out two sets of RIP updates to the directed subnet broadcast 
 AF> addresses of its interface, namely 150.150.1.255, and 150.150.2.255. 
 AF> 
 AF> Router2 will process the RIP packets destined to the 150.150.1.255 
 AF> subnet broadcast. The question is what should it do with the 
 AF> RIP packets sent by Router1 to the 150.150.2.255 subnet broadcast? 
 AF> 
 AF> a) drop the rip packets as these packets are mac layer broadcasts 
 AF>    and therefore should not be routed. 

My understanding of RIP is that they are never routed anyway.  They are processed if valid.  The only RIP packets that one should see are updates (or responses) originated from the sending router (or host).  The RFC, 1058, I think, does not mention multiple subnets but I'm told there is an RFC that covers this (rfc #?? anyone). 

 AF> b) route the packets back to 150.150.1.1 since this router knows 
 AF> how to get to the 150.150.2.0 subnet. 

I've seen a mixed router environment consisting of cisco 8.0x (old I 
know) and Xyplex 3.1 (also old) do this whenever the cisco was plugged into the net.  Also, the cisco 8.0x did not correctly stop doing this once the other subnets were defined to it.  Upgrading to cisco 9.xx 
solved the problem and infact included better support for defining 
secondary IP addresses for an interface. 

I cannot remember if it was in the cisco or Xyplex docs but one of those stated that ALL routers on the same broadcast network must have addresses defined in ALL subnets lest routing loops be created. This seemed to be bourne out in my Xyplex/cisco experience. 

Now however, the mix has grown with the addition of a IBM 6611.  So 
far, only one subnet is defined to the 6611 and no loops.  Unfortunately, the 6611 does not know how to route to subnets other than the one 
defined to it.  Haven't figured that out.  It won't add the other 
subnets to it's routing table - maybe out of self-preservation.  Trying to use the old cisco way of defining static routes to the other subnets met with some problems attempting to apply the configuration update.  So it never got a good test and now it's in production. 

Moral of all this as always "It depends" 

What routers/software rels. were you using? 

Oh.. and one final note... Xyplex acts as your router1 sending updates for each subnet defined but not necessarily identical updates - curious, btw. cisco 9.xx seems to only send one update for each net - seemly the primary IP address in each net.  In my case I was migrating from a class C net addr borrowed from a vendor for testing to the registered class B net addr.  


-----------[000046][next][prev][last][first]----------------------------------------------------
Date:      3 May 1994 13:50:43 GMT
From:      sarr@citi.umich.edu (Sarr J. Blumson)
To:        comp.protocols.tcp-ip
Subject:   Re: OSF DCE licensing

In article <2q3p68INN3ud@dur-news.ctron.com>, slama@ctron.com (Frederick G.
Slama) writes:
|> 
|> 	How?
|> 
|> 	I would like to acquire the DCE developement tools for SunOS4.1.3.
|> 	Is this possible? And at what cost?

Send mail to sales@transarc.com.

-- 
--------
Sarr Blumson                         sarr@citi.umich.edu
voice: +1 313 764 0253               home: +1 313 665 9591
CITI, University of Michigan, 519 W William, Ann Arbor, MI 48103-4943

-----------[000047][next][prev][last][first]----------------------------------------------------
Date:      3 May 94 19:06:06
From:      drw@zermelo.mit.edu (Dale R. Worley)
To:        comp.os.ms-windows.networking.tcp-ip,comp.os.os2.networking.tcp-ip,comp.protocols.tcp-ip.ibmpc,comp.protocols.tcp-ip,comp.sys.ibm.pc.misc
Subject:   Remote magnetic tape protocol *server* wanted for PC

I'm looking for software that I can load on a PC that will function as
a *server* for remote magtape accesses, according to the usual BSD
"rmt" protocols.

Oh, yes, and while we're at it, what's a good way of writing a tar
file to a tape on a remote machine?

Please e-mail responses; I'll post a summary.

Dale

Dale Worley		Dept. of Math., MIT		drw@math.mit.edu
--
The good news is that we have the entire human genome in the computer.
The bad news is that the computer alphabetized it.

-----------[000048][next][prev][last][first]----------------------------------------------------
Date:      Tue, 3 May 1994 15:00:15 GMT
From:      mark@alias.com (Mark Andrews)
To:        comp.unix.internals,comp.unix.misc,comp.unix.bsd,alt.books.technical,comp.protocols.tcp-ip,comp.unix.programmer
Subject:   Re: Book on BSD's network implementation

If you want the BSD very specific implementation, look at the 4.3BSD
Networking Implementation Notes in the System Manager's manual.

-----------[000049][next][prev][last][first]----------------------------------------------------
Date:      Tue, 3 May 1994 15:37:02 GMT
From:      jantypas@netcom.com (John Antypas)
To:        comp.protocols.tcp-ip
Subject:   Could someone update me on IP futures?

It's been awhile since I've tracked IPv4.  With the world now going IP
(at least the US), the address we feared may soon be upon us.  I know
there were several solutions proposed to extend the address space.
Could some kind person list them and provide a summery or just references
for them?
-- 
John Antypas / 21st Century Software Walnut Creek CA
jantypas@soft21.s21.com


-----------[000050][next][prev][last][first]----------------------------------------------------
Date:      3 May 1994 16:17:54 GMT
From:      wdp@roadnet.ups.com (Bill Paris)
To:        comp.protocols.tcp-ip
Subject:   Network Time Synchronization

I'm seeking information on network clock synchronization.

Our goal is to implement a mechanism for synchronizing the clocks
of the computers (non-Unix) in our TCP/IP network.  We'll use the
traditional approach: one computer is assigned master time keeper
status, the others reset their clocks according to some protocol.

I've read several Internet RFCs, in particular RFC 958 "Network Time 
Protocol (NTP)" by D.L. Mills, and have obtained the software for the 
BSD Time Synchronization Protocol "timed" server.  My questions are 
as follows:

1. We see our choices as being either porting BSD's timed to OS/2 or 
   implementing NTP from scratch.  Are there other options I'm unaware
   of?  Any commercial products?

2. BSD's timed uses IP's ICMP timestamp messages, but Mills in RFC 956
   "Algorithms for Synchronizing Network Clocks" says that support for
   these messages in the DoD Internet protocol suite is optional, and
   not generally supported.  Is this true?  Does this imply that timed
   is of limited use?

Thanks in advance,
Bill Paris





-----------[000051][next][prev][last][first]----------------------------------------------------
Date:      3 May 1994 16:29:17 GMT
From:      barmar@think.com (Barry Margolin)
To:        comp.protocols.tcp-ip
Subject:   Re: Multiple masks on a single machine

In article <075303Z01051994@anon.penet.fi> an94584@anon.penet.fi writes:
>IP ADDRESS              SUBNET MASK
>== =======              ===========
>aa.bb.250.100           FF.FF.00.00
>aa.bb.128.1             FF.FF.80.00
>aa.bb.32.1              FF.FF.E0.00
>
>Will that allow me to access the aa.bb.128.0 sunet over the 128.0 interface
>the 32.0 subnet over the 32.0 interface and the remainder over the 
>250.0 interface

No.  It will also try to access subnets aa.bb.33-47.0 over the 32.1
interface, since all those addresses match when you apply that subnet mask.

Set the masks of the 32.1 and 128.1 interfaces to FF.FF.FF.00.  This way,
only an exact match on the third octet will cause those interfaces to be
used.
-- 
Barry Margolin
System Manager, Thinking Machines Corp.

barmar@think.com          {uunet,harvard}!think!barmar

-----------[000052][next][prev][last][first]----------------------------------------------------
Date:      3 May 94 22:33:44 MDT
From:      jrd@cc.usu.edu (Joe Doupnik)
To:        comp.protocols.tcp-ip
Subject:   Re: NetWatch

In article <WZEOBTLG@math.fu-berlin.de>, schaumann@oms2.usmc.mil (David K Schaumann) writes:
> Someone had posted a message about the freeware package
> called NetWatch.  Now, how do you configure the netdev.sys
> file?  Will this work inconjuction with NDIS drivers?
> 
> Thanks,
> 
> Dave
> 
> David Schaumann
> Okinawa    Japan
> InterNet: schaumann@oms2.usmc.mil
> or: schaumann@okr-smtp1.usmc.mil
----------
	Netwatch requires that netdev.sys be loaded but otherwise the
contents are not used. Just leave it intact or run Custom against it.
	Netwatch uses Ethernet Packet Drivers in promiscuous mode. It
does not speak NDIS, though you can try NDIS to PD shim dis_pkt9
(anonymous ftp to netlab2.usu.edu, cd drivers for dis_pkt9.zip; anon
ftp to netlab1.usu.edu, cd pub/netwatch, or to netlab.usu.edu, cd
[.netwatch] for netwatch).
	Joe D.

-----------[000053][next][prev][last][first]----------------------------------------------------
Date:      Tue, 3 May 1994 16:37:19 GMT
From:      atkinson@itd.itd.nrl.navy.mil (Ran Atkinson)
To:        comp.protocols.tcp-ip
Subject:   Re: IP Addresses and the Internet?

In article <2q4vek$j9m@cronkite.cisco.com> tli@cisco.com (Tony Li) writes:

>I suggest you look at RFC 1597 as it deals with most of these questions.
>
>Tony

  Folks should note several things that Tony omitted (and IMHO should
have mentioned for completeness):

	(1) RFC-1597 represents the opinions of several individuals,
		and was not reviewed by the IETF, IESG, or IAB.
	(2) RFC-1597 is NOT on the standards-track.  It is only an
		Informational RFC.
	(3) A significant part of the Internet community strongly disagrees
		with many of the assertions and conclusions of RFC-1597.
	(4) An Internet Draft rebutting RFC-1597 is in preparation by
		several different members of the Internet community.

My own quick analysis of RFC-1597 led me to find what I think is a new
and significant potential security attack arising when a typical
firewall is used with networks configured as RFC-1597 suggests.  
I haven't confirmed the existence of the attack yet myself, so I
deliberately use the word "potential" here.

Ran
atkinson@itd.nrl.navy.mil



-----------[000054][next][prev][last][first]----------------------------------------------------
Date:      3 May 1994 16:40:53 GMT
From:      barmar@think.com (Barry Margolin)
To:        comp.protocols.tcp-ip
Subject:   Re: Destination Unreachable

In article <2prp9n$8il@cmcl2.NYU.EDU> roy@mchip00.med.nyu.edu (Roy Smith) writes:
]	When I ping from 205.115 to 205.14, I get Desination Unreachables
]and I don't understand why (see ping output below).  One came from 127.0.0.1
](i.e.  the loopback interface), and others from the two SGI's (205.96 and
]135.9).  What does this mean?  What I expected was to just not get any
]response from the powered-off host, and can't figure out what the Dest
]Unreachables mean and why I'm getting them.

I don't think those packets are in response to your ping.  In verbose mode,
ping shows *all* ICMP packets that the host receives.

]mchip00> ping -v 128.122.205.14
]PING 128.122.205.14: 56 data bytes
]36 bytes from 127.0.0.1: icmp_type=3 (Dest Unreachable)
]x00: x00240045
]x04: x00003fd8
]x08: x000001ff
]x0c: x0100007f
]x10: x0100007f
]x14: xc1b20303
]x18: x00000000
]x1c: x007b0045
]x20: x00003fd7
]x24: x0000111e
]x28: x0100007f
]x2c: x0100007f
]icmp_code=3

icmp_code=3 is "Port unreachable".  The above ICMP packet is for a UDP
packet (octet 0x26 is 0x11 = UDP) that was sent to an unbound port.  This
was happening at the same time as your ping, so it showed up in the output.
I suspect all the other ICMP errors were similar.
-- 
Barry Margolin
System Manager, Thinking Machines Corp.

barmar@think.com          {uunet,harvard}!think!barmar

-----------[000055][next][prev][last][first]----------------------------------------------------
Date:      3 May 1994 16:56:54 GMT
From:      huitema@mitsou.inria.fr (Christian Huitema)
To:        comp.protocols.tcp-ip
Subject:   Re: IP Addresses and the Internet?

In article <2q5aju$7r7@erinews.ericsson.se>, etxmesa@eos.ericsson.se (Michael Salmon) writes:
|> In article <herbrCp7FnC.JME@netcom.com>
|> herbr@netcom.com (Herb Rosenberg) writes:
|>
|> |> Is there any RFC's that talk about firewalls specifically?
|> 
|> None that I know of, the closest I could quickly find was 1579.
|> 

You may want to look at the internet draft
"draft-iab-sec-arch-workshop-00.txt" which reports the result of a recent IAB
sponsored workshop on "Security and the Internet Architecture". This draft is
still being edited, the final version should soon appear as an informational
RFC. 

As for your initial question on "how many addresses do I need", I suggest you
play safe and get coordinated addresses for your whole net. The cost is
minimal and the nightmares you could get otherwise may be quite interesting.
Just suppose you buy another company or that you decide to establish open
connectivity with some of your providers. Or even that Internet security
suddenly increases to the point where you don't need the firewall at all :-)

Christian Huitema

-----------[000056][next][prev][last][first]----------------------------------------------------
Date:      3 May 1994 18:50:12 GMT
From:      sean@oak.his.ucsf.edu
To:        comp.protocols.tcp-ip,comp.dcom.sys.cisco
Subject:   multiple gateways

Situation: multiple LANs interconnected via cisco routers and T1s.

Problem:  most hosts (PCs, Macs, terminal servers) only know about one
'gateway' off the local net.  If a router dies the hosts don't know
about another one on the same LAN.  From the end user perspective there
is no redundancy.

Solutions?

Run a gateway discovery protocol.  Not available on most non unix hosts.

Configure the hosts with a 0.0.0.0 netmask and let the routers do proxy
ARP.  The hosts will lose connections when a router dies, but can find a
new gateway by bouncing.  Any reasons not to do this?

Any other suggestions?

Thanks

Sean McGrath - Voice: 510.653.8387
Email: sean@oak.his.ucsf.edu
Paper: 638 66th st, Oakland Ca, 94609

-----------[000057][next][prev][last][first]----------------------------------------------------
Date:      3 May 1994 19:03:45 GMT
From:      learned@winternet.mpls.mn.us (Ed Learned)
To:        comp.protocols.tcp-ip
Subject:   TCP/IP & AS/400

I am interested in establishing a TCP environment on my AS/400. Any
tips, warnings, suggestions, etc?

--

    Ed Learned                              |  Information
    Ed.Learned@mmbbs.mn.org                 |  Highway
    learned@.winternet.mpls.mn.us           |  Worker

-----------[000058][next][prev][last][first]----------------------------------------------------
Date:      3 May 1994 20:44:00 GMT
From:      sac@oti-hsv.com (Scott Cechovic)
To:        comp.protocols.tcp-ip
Subject:   tip with cslip problems

[ Article crossposted from comp.protocols.ppp ]
[ Author was Scott Cechovic ]
[ Posted on 3 May 1994 20:42:40 GMT ]

I am having a problem getting the version of tip that came with cslip-2.7 to
recognize when the modem loses the carrier.  As a result, when the connection
is lost tip never terminates.  The local modem is a Telebit t3000 with the 
following settings:
T3000SA - Version LA7.02 - Active Configuration
 B1  E1  L1  M1  T   Q0  V1  X12  Y0
&C1 &D3 &G0 &J0 &L0 &Q0 &R2 &S1 &T4 &X0
S000=0   S001=0   S002=43  S003=13  S004=10  S005=8   S006:3   S007=60
S008=2   S009=6   S010=14  S011=70  S012=50  S018=0   S025=5   S026=1
S038=0   S041=0   S045=0   S046=0   S047=4   S048=0   S050=0   S051:6
S056=17  S057=19  S058:0   S059=15  S060=0   S061=1   S062=15  S063=0
S064=0   S068:2   S069=0   S090=0   S093=8   S094=1   S100=0   S104=0
S105=1   S111=255 S112=1   S113=126 S114=0   S115=0   S151=4   S155=0
S180=2   S181=1   S183=25  S190=1   S253=10  S254=255 S255=255

Any suggestions??
Thanks,
Scott


--

+===============================()===================================+
|                                 0                                  |
|               ()      +--------. o   Scott A. Cechovic             |
|                () ---. \ ) ) *  \ *  Optimization Technology, Inc. |
|     +--------.  0    |>-> ) ) )  o   5021 Technology Dr., Suite 1A |
| ---. \ ) ) *  \ *   -' / ) ) )  /    Huntsville, AL 35805          |
|    |>-> ) ) )  o      +--------'     scott.cechovic@oti-hsv.com    |
|   -' / ) ) )  /                      (205) 721-1288                |
|     +--------'                                                     |
+====================================================================+

--

+===============================()===================================+
|                                 0                                  |
|               ()      +--------. o   Scott A. Cechovic             |
|                () ---. \ ) ) *  \ *  Optimization Technology, Inc. |
|     +--------.  0    |>-> ) ) )  o   5021 Technology Dr., Suite 1A |
| ---. \ ) ) *  \ *   -' / ) ) )  /    Huntsville, AL 35805          |
|    |>-> ) ) )  o      +--------'     scott.cechovic@oti-hsv.com    |
|   -' / ) ) )  /                      (205) 721-1288                |
|     +--------'                                                     |
+====================================================================+

-----------[000059][next][prev][last][first]----------------------------------------------------
Date:      Tue, 3 May 1994 21:06:26 GMT
From:      jarocki@dvorak.amd.com (John Jarocki)
To:        comp.protocols.tcp-ip
Subject:   Re: netmask to reduce class C Network (?)

In article <588NBMLS@cumin.telecom.uqam.ca>,
 <optel@phoque.info.uqam.ca> wrote:
>
>Can I use a netmask to reduce the network portion of a class C address?
>I have 2 class C addresses 193.220.250 and 193.220.251 each currently supporting
>about 200 users and would like to rehome them into an Ethernet Switch and 
>therefore stop using a router between them.
>
>I have:
>    <--------- net ----------> <-host->
>    NNNNNNNN NNNNNNNN NNNNNNNN HHHHHHHH
>
>I want:
>    <-------- net ---------><---host-->
>    NNNNNNNN NNNNNNNN NNNNNNHH HHHHHHHH 

Fascinating question.  We did a very similar thing
in one our networks when we installed an ethernet switch.
Actually, we changed from class C to class B and let
our Cisco AGS+ router handle proxy arp response for
the networks that were on the other side of the router.

Bad idea for two reasons.  One was performance.  For
reasons that we still have not completely figured out,
this made interactive network traffic very slow.  I can
only imagine because the router was having to do a lot
of work.  The second (and more important reason) was that
the group we did this for neglected to mention that one
of the workstations on this network was a gateway for
an Apollo token ring.  Well, the router couldn't very
well service arps for that network -- it didn't really
even know about it -- except that that there was a
gateway.

Anyway, to make a long story short.... ;-)

What we ended up doing was adding some static routes
on each machine connected to the switch.  We were
merging 163.181.18 and 163.181.14, so we set up route
entries like this:
beast [1]% netstat -r
Routing tables
Destination      Gateway            Flags     Refs     Use  Interface
localhost        localhost          UH          0 27168066  lo0
default          funny18nw          UG          3    25327  lan0
163.181.18       beast              U          41  3745055  lan0
163.181.14       beast              U          17  1761038  lan0

Now 18 net machines arp for 14 net machines and vice-versa.

It works quite well (except that apollos running Domain/OS will
not allow you to set a static route entry with 0 hops for
a non-local network, but that's another story).

If that's not clear, send me an email.

>Thank you for helping me out 
>
>Marc Louarduzzi
>opTEL@phoque.info.uqam.ca 

regards,
--john


-- 
________________________________________________________________________
 John Jarocki         Systems Engineer/Administrator
 _______
 \____  |Advanced     CAD Technology & Systems
 /|   | |Micro        5204 E. Ben White M/S 553
| |___| |Devices      Austin, Texas 78741
|____/ \|
---------john.jarocki@AMD.COM voice:512-602-4098 fax:512-602-5156-------

-----------[000060][next][prev][last][first]----------------------------------------------------
Date:      Tue, 3 May 1994 23:18:53 GMT
From:      schaumann@oms2.usmc.mil (David K Schaumann)
To:        comp.protocols.tcp-ip
Subject:   NetWatch

Someone had posted a message about the freeware package
called NetWatch.  Now, how do you configure the netdev.sys
file?  Will this work inconjuction with NDIS drivers?

Thanks,

Dave

David Schaumann
Okinawa    Japan
InterNet: schaumann@oms2.usmc.mil
or: schaumann@okr-smtp1.usmc.mil

-----------[000061][next][prev][last][first]----------------------------------------------------
Date:      4 May 1994 09:36:07 -0600
From:      thayne@xmission.com (Thayne Forbes)
To:        comp.protocols.tcp-ip
Subject:   Re: Bargains on Comer's 3 Volume TCP/IP set

Michael Saunby (swssauby@reading.ac.uk) wrote:

:> >:> Scott Austin (scotta@kije.gsfc.nasa.gov) wrote:
 
:> >:> : The Library of Computer and Information Sciences (LCIS), is currently
:> >:> : having a promo with Comer's 3 books.  You can save some bucks, since the
:> >:> : Publisher's prices on these is $164!  This offer is open to any
 
:> $164!!  I guess that is why my "Unabridged paperback reprint" has
:> "This edition may be sold only to those countries to which it is
:> consigned by Prentice Hall International.  It is not to be re-exported,
:> and is not for sale in the U.S.A., Mexico, or Canada." printed on the cover.
 
:> I only have volume 3 which cost me 22.95 pounds (UK), and I thought that
:> was expensive!

To be fair, I have used the PHI reprints and they are nowhere near the
quality of the originals.  The originals are hardback and the paper
quality is much better.  You get what you pay for.

-- 
Thayne Forbes           thayne@xmission.com
Computer Weenie at large

-----------[000062][next][prev][last][first]----------------------------------------------------
Date:      Wed, 4 May 1994 00:48:40 GMT
From:      atkinson@itd.itd.nrl.navy.mil (Ran Atkinson)
To:        comp.protocols.tcp-ip
Subject:   Re: Network Time Synchronization

In article <2q5tfi$epo@news.roadnet.ups.com> wdp@roadnet.ups.com (Bill Paris) writes:

>I've read several Internet RFCs, in particular RFC 958 "Network Time 
>Protocol (NTP)" by D.L. Mills, and have obtained the software for the 
>BSD Time Synchronization Protocol "timed" server.  My questions are 
>as follows:

  It is generally wise to obtain an rfc-index and work backwards in
case older RFCs have been replaced by updated versions.  In this case,
RFC-958 is long deprecated.  I think RFC-1305 is reasonably current,
but again it is wise to check the rfc-index file (available at the
same exact place rfcs are kept online :-).

>1. We see our choices as being either porting BSD's timed to OS/2 or 
>   implementing NTP from scratch.  Are there other options I'm unaware
>   of?  Any commercial products?

Why not use the freely distributable NTPv3 implementation available
from louie.udel.edu ??  

For details on NTP, go read comp.protocols.time.ntp.  Timed is
unlikely to work as you'd like it to.  NTP almost certainly will get
the job done.

Ran
atkinson@itd.nrl.navy.mil

Followups are suitably redirected.

-----------[000063][next][prev][last][first]----------------------------------------------------
Date:      Wed, 4 May 1994 01:57:21 GMT
From:      wagoner@zorro.wrc.xerox.com (William Wagoner)
To:        comp.protocols.tcp-ip
Subject:   Broadcast storm w/pppd2.04 & Sun4.1.3

I'm having a problem linking a remote Sun to our network using
pppd 2.04.  I'm running SunOS 4.1.3 on both ends.  The ppp_driver.o
is being modloaded properly, and our ethernet broadcast address
is set to all 1s, as are all other systems on our net
(broadcast addr 13.0.35.255, subnet mask 255.255.252.0, net
13.0.34.0).
My work machine started participating in the broadcast storms after I
began to leave it with the drivers loaded and had connected from the
remote machine.  At first we discovered that in.routed was running 
on the work machine, and thought it was messing up the routes. But 
after setting up a static route at work, the problem continued.

We had the same problem on another system with 2 ethernet adapters
and on one with a FDDI and ethernet adapter.  In both cases, the 
kernel was changed to set ipforwarding off (_ip_forwarding?W-1 
for you adb fans).  This stopped the broadcast storms, but prevents
the remote machine from seeing into the local net past the work 
machine (no X, news, www, etc, except by telneting to the work host).
Once the problem starts, modunloading the ppp driver 
won't stop it from happening.  Only a reboot or turning ipforwarding 
off seems to work.  I can't believe that Sun's TCP/IP only works
properly with 1 interface, so I'm hoping someone else out there
has run into this before, and will have a suggestion.

Feel free to email me, and I'll summarize.

Thanks!!

Bill Wagoner
Xerox Webster Research Center
Rochester, NY
(716)422-5754
wagoner@wrc.xerox.com

#include <std_disclaimers.h>






-----------[000064][next][prev][last][first]----------------------------------------------------
Date:      4 May 1994 03:00:45 GMT
From:      heller@akh104.rh.psu.edu (Andrew K. Heller)
To:        comp.protocols.tcp-ip
Subject:   SLIP/PPP info NEEDED!!!!!

  There is a project on the move that needs this information..for
all tense and purposes ... assume I am trying to add SLIP TCP/IP
to a machine other than PC/MAC...a machine that doesnt have  much
memory...   

  So..where would I look on developing a driver and related tools?
i.e. where are the manuals on the protocol and examples?

--
Andrew K. Heller - Making the world safe once again.
------------------------------------------------------------------
Welcome to P$U, please bend over.

-----------[000065][next][prev][last][first]----------------------------------------------------
Date:      Wed, 4 May 1994 03:14:38 GMT
From:      vjs@calcite.rhyolite.com (Vernon Schryver)
To:        comp.protocols.tcp-ip
Subject:   Re: Network Time Synchronization

In article <2q5tfi$epo@news.roadnet.ups.com> wdp@roadnet.ups.com (Bill Paris) writes:
>...
 
>2. BSD's timed uses IP's ICMP timestamp messages, but Mills in RFC 956
>   "Algorithms for Synchronizing Network Clocks" says that support for
>   these messages in the DoD Internet protocol suite is optional, and
>   not generally supported.  Is this true?  Does this imply that timed
>   is of limited use?

Years ago I wrote something to synchronize a corporate network to a
neighbor's system with a "satelite clock" over a 9600 bps "cypress"
link.  It uses ICMP time stamps for the time and port 13 for the date.
It made its way into a workstation vendor's product, and is used by a
small fraction of that company's customers but a substantial number in
absolute terms.  (Since timed is shipped turned on by default, most of
those customers use timed.)  I occassionally but very rarely hear of
places where it doesn't work but only because port 13 is not supported,
never because ICMP timestamps don't work.  Therefore, I think Dave Mills'
statement in the ancient (from 1985) RFC-956 has become wrong.


Check the more recent RFC-1361 from 1992.  Look on louie.udel.edu for
NTP source.  Read comp.protocols.time.ntp.

If you want my code, which you probably don't since NTP would probably
be a better choice, look for "timeslave" in the slightly post-4.4BSD
version of timed source on ftp.sgi.com.


Vernon Schryver    vjs@rhyolite.com

-----------[000066][next][prev][last][first]----------------------------------------------------
Date:      4 May 1994 12:11:19 -0400
From:      lizw@cs.umd.edu (Elizabeth White)
To:        comp.sys.hp.hpux,comp.protocols.tcp-ip
Subject:   Traffic generation software needed


Does anyone know where I can get some TCP/IP traffic generation software called
"npc" or some other traffic generation software (not spray) that runs on an 
HP 9000 system? I need to be able to vary network loading parameters, 
size of packets, types of packets, etc.  What would be best for me would 
be a package that has a client and a server piece that would allow me to 
send traffic and monitor how much arrived and at what rate.  Any info will 
be greatly appreciated.  Thanks.


Liz White
lizw@cs.umd.edu
Computer Science Dept
University of Maryland, College Park


-----------[000067][next][prev][last][first]----------------------------------------------------
Date:      Wed, 04 May 1994 14:28:28 -0500
From:      johns@oxygen.house.gov (John Schnizlein)
To:        comp.protocols.tcp-ip
Subject:   Re: osi vs tcp/ip

In article <2q8mm4$819@gopher.cs.uofs.edu>, jpb10@owl.cs.uofs.edu (James
Barrett) wrote:

>   I am trying to contrast OSI (open systems interconnection) with TCP/IP.
> However, I am having difficulty understanding OSI addressing and routing
> schemes.  I know the basic TCP address such as 134.198.11.10, but I need
> an example of an OSI address.
> 
You are so brave to post this kind of question in this newsgroup that you
deserve an answer (not just RTFM). This summary is simplified. OSI bigots
seem to love the complexity required of a complete answer. IP bigots seem
to love to scoff at this love of complexity. Too bad this bigotry colors
the scene.

OSI addresses are called Network Service Access Points (NSAPs). One of the
glories of NSAPs is that they can be from 12 to 20 bytes (called octets in
OSI-speak) long, depending on the Authority and Format Identifier (AFI),
which is the first byte, and the Domain (specific part) Format Identifier
(DFI), which is the 4th byte of the address.

The most popular OSI NSAP form, comparable to an IP address, is the NSAP
structure specified in GOSIP-2. It is 20 bytes long and is designed for
multiple administrative layers of allocation authority. This GOSIP-2 NSAP
is ideal for connectionless packet networks like the Internet, which needs
more layers of address allocation authority. The network (layer-3) protocol
equivalent of IP is called CLNP (ConnectionLess Network Protocol).

NSAPs are written in hexadecimal rather than dotted decimal.
Take a deep breath now, we are about to explore the fields of an NSAP
(address).
AFI (see above) is byte 1. For GOSIP-2 its value is 47 (remember hex now).
IDI (Initial Domain Identifier) is bytes 2 & 3. Its GOSIP-2 value is 00 05.
DFI (see above) is byte 4. Its GOSIP-2 value is 80.
AA (Administrative Authority) is bytes 5 - 7. You get this assigned to your
org.
   This is comparable to an Autonomous System number in the Internet.
Bytes 8 & 9 are reserved.
RD (Routing Domain) is bytes 10 & 11. This is the first layer of address   
   
   structure available to the organization for assignment. Notice that the
size
   of this field is comparable to all the class-B network numbers.
Area is bytes 12 & 13. This is the second layer of addressing within the
org.
ID is bytes 14 - 19. This is any unique end system identifier.
   MAC (IEEE 802) addresses are just this size and work well.
Sel (selector) is byte 20, identifing the user process within the end
system.
    The selector is comparable to the protocol field in an IP header.

OK, now here is an example. This is the NSAP of my OSI router:
47.0005.8000.5800.0000.0001.0001.0000.0C04.4C58.00
-- 
Badges! we don't need no stinking badges!         |
disclaimer! we don't need no stinking disclaimer! | John M. Schnizlein
everybody knows nobody can represent the views of | johns@oxygen.house.gov
435 elected policy makers.                        | router jockey

-----------[000068][next][prev][last][first]----------------------------------------------------
Date:      Wed, 4 May 1994 08:27:08 GMT
From:      swssauby@reading.ac.uk (Michael Saunby)
To:        comp.protocols.tcp-ip
Subject:   Re: Bargains on Comer's 3 Volume TCP/IP set

thayne@xmission.com (Thayne Forbes) writes:

>Cliff Bedore (cliffb@skeeter.bdb.com) wrote:
>:> Scott Austin (scotta@kije.gsfc.nasa.gov) wrote:
>:> : Hi,
 
>:> : The Library of Computer and Information Sciences (LCIS), is currently
>:> : having a promo with Comer's 3 books.  You can save some bucks, since the
>:> : Publisher's prices on these is $164!  This offer is open to any

$164!!  I guess that is why my "Unabridged paperback reprint" has
"This edition may be sold only to those countries to which it is
consigned by Prentice Hall International.  It is not to be re-exported,
and is not for sale in the U.S.A., Mexico, or Canada." printed on the cover.

I only have volume 3 which cost me 22.95 pounds (UK), and I thought that
was expensive!

-- 
	Michael Saunby (M.Saunby@reading.ac.uk) 
	Meteorology Dept, Reading University, UK


-----------[000069][next][prev][last][first]----------------------------------------------------
Date:      Wed,  4 May 94 19:23:00 -0600
From:      xerxes@bgbbs.com (Xerxes)
To:        comp.protocols.tcp-ip
Subject:   OS/2 DNS failure

I am using the OS/2 TCP/IP DNS as a primary DNS.  It works for a 
while, but eventually locks up.  This only happens on kava.nisc.sri.com
If this entry is removed from named.ca, the DNS still eventually locks
up, when kava is accessed by/from another name server.

Is anyone else using the OS/2 DNS as a primary nameserver?  Is anyone
using the OS/2 DNS as caching? Is anyone using the OS/2 DNS at all?

And if you are, do you have problems with kava too?

            Xerxes

---
þ CmpQwk #UNREGþ UNREGISTERED EVALUATION COPY


-----------[000070][next][prev][last][first]----------------------------------------------------
Date:      4 MAY 94 10:42:48 GMT
From:      carroll@pppl.gov (Tom Carroll)
To:        comp.protocols.tcp-ip
Subject:   NFS-problems on VMS (to HP/E&S)


	NFS VMS-HP problems

We have problems running a VMS machine (op 5.5-2) as server and a HP735 (and
Evans & Sutherland ESV3+) as client using NFS.

We have just installed DEC TCP/IP Service for OpenVMS Version 2.0D on the
VMS machine.  Telnet and ftp works fine, both in and out.  But, when trying to 
mount the VMS-disk on the HP and ESV we get the error message:

mount: rabbit:/rabbit1 server not responding: RPC: Authentication error; why = 
Invalid client credential

Does anyone have any idea what is wrong?

The Hp is running HP-UX 9.03

Thanks on beforehand!

Anders Svensson
E-mail:	Anders.Svensson@mbfys.lu.se

-----------[000071][next][prev][last][first]----------------------------------------------------
Date:      Wed, 4 May 1994 11:20:36 GMT
From:      kkchin@dcs.warwick.ac.uk (K K Chin)
To:        comp.protocols.tcp-ip
Subject:   SLIP implementation on SUN OS

can anyone please help to answer some question on the SLIP implementation on sun os :

1.	Is SLIP actually implemented ? If yes is it a standard component or an option ?

2.	Will SLIP filter out the IP packet that went out the serial port, since only one host will be connect to that port ? Say if a SLIP interface is setup to connect to host A will the packet for host B be sent out through this port (and leave the SLIP interface at host A to filter out the unwanted packet) ?
-- 
this is a file

-----------[000072][next][prev][last][first]----------------------------------------------------
Date:      Wed, 4 May 1994 16:33:28
From:      roberto@jsb.co.uk (Robert Osborne)
To:        comp.protocols.tcp-ip
Subject:   Telnet Echo Option

This is the first time that I have ever sent anything, so if I am doing this 
wrong I apologise.

Can anybody help me with a problem I am having with the Telnet echo option. If 
I turn off all echoing, local and remote, everything appears to be fine upon 
the initial login. No characters are echoed at all, as I expect to happen. 

However, once the login process has been completed, and I am at the usual 
command line prompt, whatever is typed by the user is echoed back. I have not 
received any command telling me that remote echoing has been turned back on.

Is this is a standard feature, or something wrong with my remote host, or have 
I completely misunderstood the echo option.

Rob Osborne

-----------[000073][next][prev][last][first]----------------------------------------------------
Date:      Wed, 04 May 1994 20:29:00 -0500
From:      Bill.Beers@f240.n379.z1.fidonet.org (Bill Beers)
To:        comp.protocols.tcp-ip
Subject:   Re: Variable IP subnets with RIP on Wellfleet and Cisco routers

 A few days ago, Paul Ferguson (paul@sprintlink.net) wrote: 

 PF> Actually, what I MEAN, is that OSPF is geared towards networks in which 
 PF> transferrence of the subnet mask is desirable (variably subnetted 
 PF> networks eg.). Most of the driving factors in implementing OSPF are 
 PF> specialized networks, where a specific routing requirements exist. 

One benefit that OSPF (at least from Xyplex) gives is unnumbered 
WAN links.  With RIP, each link eats a subnet, which contributes 
to the IP address depletion. 

Have a question about OSPF:  Does it allow multiple subnets to exist 
on the same broadcast network?  Seems to me I've read that OSPF does 
not allow this.  Thus, one may be forced to implement variable subnet masks to allow some parts of the net to have more host numbers than 
others.  This would seem to introduce a need for a structured design 
for subnet address assignments.  But if OSPF would tolerate multiple 
subnets, then one could continue with a fixed subnet mask and sequential subnet address assignment.  Guess that's the question. 



-----------[000074][next][prev][last][first]----------------------------------------------------
Date:      Wed, 4 May 1994 13:50:02 GMT
From:      ihsan@nmti.com (jaleel ihsan)
To:        comp.protocols.tcp-ip,comp.dcom.sys.cisco
Subject:   Re: multiple gateways

In article <2q66d4$evl@agate.berkeley.edu> sean@oak.his.ucsf.edu writes:
>Situation: multiple LANs interconnected via cisco routers and T1s.
>
>Problem:  most hosts (PCs, Macs, terminal servers) only know about one
>'gateway' off the local net.  If a router dies the hosts don't know
>about another one on the same LAN.  From the end user perspective there
>is no redundancy.
>
>Solutions?
>
>Run a gateway discovery protocol.  Not available on most non unix hosts.
>
>Configure the hosts with a 0.0.0.0 netmask and let the routers do proxy
>ARP.  The hosts will lose connections when a router dies, but can find a
>new gateway by bouncing.  Any reasons not to do this?

Cant do this if a temporary break in the connection is unacceptable.

>
>Any other suggestions?

I am looking for a better solution too !

-----------[000075][next][prev][last][first]----------------------------------------------------
Date:      Wed, 4 May 1994 15:07:33 GMT
From:      landin@cherokee.nsuok.edu (Mark Landin)
To:        comp.protocols.tcp-ip,comp.sys.mac.comm,comp.sys.novell
Subject:   Re: BootP on Novell's NLM and MacTCP

opie@panix.com (Matthew S. "Opie" Warren) writes:

[Story of non-working bootp symptoms deleted]

>What's up with this?  The only difference I can think of is that in the
>non-working site, the BootP server is in a different subnet from the
>requesting Macs.  Our networking people, however, assure me that this would
>not matter.

Well, you might need to set up a static route in the "GATEWAY" file 
that BOOTP uses to get across the subnet. I forget exactly what the
statement should look like, but I think that's what you need to do.

-- 
*-----------------------------------------------------------------------------*
*  Mark C. Landin					Northeastern St. Univ *
*  landin@cherokee.nsuok.edu					Tahlequah, OK *
*   "Living in the pools, they soon forget about the sea" - Neil Peart, RUSH  * 

-----------[000076][next][prev][last][first]----------------------------------------------------
Date:      Wed, 4 May 1994 15:51:48 GMT
From:      rblair@cabell.vcu.edu (Richard E. Blair)
To:        comp.protocols.tcp-ip
Subject:   DNS and OpenView


Is anyone out there running HP OpenView on a Name Server?  We are
running OpenView 3.2 on an HP 715/50 running HP-UX 9.01.  Whenever
I start named and then run OpenView, I get the error message "ovw:
unable to resolve "hostname" to get a license."  The name server
itself work fine, and I believe it's configured properly.  What I
don't know is exactly how OpenView attempts to resolve a hostname.

Any suggestions would be greatly appreciated.

Richard Blair
rblair@cabell.vcu.edu

-----------[000077][next][prev][last][first]----------------------------------------------------
Date:      Wed, 4 May 1994 16:10:32 GMT
From:      xtpw@blaze.trentu.ca (Peter Wood)
To:        comp.protocols.tcp-ip
Subject:   Re: TCP/IP & AS/400

In article <2q676h$3ou@blackice.winternet.mpls.mn.us>,
Ed Learned <learned@winternet.mpls.mn.us> wrote:
>I am interested in establishing a TCP environment on my AS/400. Any
>tips, warnings, suggestions, etc?

Snag the AS/400 TCP/IP config and op Red Book - GG24-3442-01

In between this and the AS/400 docs, I was actually able to bring the
TCP/IP sub-system up in a couple of hours.

Regards,

Pete
-- 
Pete Wood - Fleming College, Peterborough, ON, Canada, K9J 7B1
Phone:	705 749-5530 x5220  FAX:  705 749-5540
-------------------------------------------------------------
Garlic is to food as insanity is to art

-----------[000078][next][prev][last][first]----------------------------------------------------
Date:      4 May 1994 23:23:51 -0400
From:      amol@ms.uky.edu (Amol Deshpande)
To:        comp.protocols.tcp-ip
Subject:   Re: Telnet Echo Option

roberto@jsb.co.uk (Robert Osborne) writes:



>However, once the login process has been completed, and I am at the usual 
>command line prompt, whatever is typed by the user is echoed back. I have not 
>received any command telling me that remote echoing has been turned back on.
 
>Is this is a standard feature, or something wrong with my remote host, or have 
>I completely misunderstood the echo option.
 
>Rob Osborne


I have almost exactly the same problem. I turned off remote echo, but as soon
as I got past the login procedure, remote echo got turned on. 

I know that the server is not sending any WILL ECHO comand before echoing...

any help appreciated,
-amol
amol@ms.uky.edu
-- 
 Amol Deshpande , amol@ms.uky.edu
Do not believe in miracles, rely on them.

-----------[000079][next][prev][last][first]----------------------------------------------------
Date:      Wed, 4 May 1994 17:23:44 GMT
From:      richlove@netcom.com (Rich Love)
To:        comp.protocols.tcp-ip,comp.dcom.sys.cisco
Subject:   Does TCP uses flow control?

When connected to a host using TCP/IP is is possible to use Xon/Xoff flow
control? All of the comms programs I've seen do not have an option for flow
control when TCP is selected.


-- Rich Love                                     Carnation Software, Inc.
 ________________________________________________________________________
| MacToPic and SBMac - Macintosh to Host connectivity and file transfer  |
| for PICK, uniVerse, unix, System Builder and other host systems.       |
| Viewpoint, Wyse 50, VT101 and Prism emulations included.               |
| Demo available for download at ftp.netcom.com in pub/carnation         |
|                                                                        |
| Phone (206) 333-4288                   Internet:  richlove@netcom.com  |
|________________________________________________________________________|

-----------[000080][next][prev][last][first]----------------------------------------------------
Date:      Wed, 4 May 1994 17:36:15 GMT
From:      geertj@ripe.net (Geert Jan de Groot)
To:        comp.protocols.tcp-ip
Subject:   Re: IP Addresses and the Internet?

[Please note that I one of the authors of RFC1597]

In <Cp8JI9.67F@ra.nrl.navy.mil> atkinson@itd.itd.nrl.navy.mil (Ran Atkinson) writes:
>  Folks should note several things that Tony omitted (and IMHO should
>have mentioned for completeness):
>	(1) RFC-1597 represents the opinions of several individuals,
>		and was not reviewed by the IETF, IESG, or IAB.
>	(2) RFC-1597 is NOT on the standards-track.  It is only an
>		Informational RFC.
True. Clearly stated in the introduction. A significant amount of RFC's
are not on the standards-track. We have chosen for this approach to save
time; we had some requests pending on the actual assignment of the
address space for RFC1597.

>	(3) A significant part of the Internet community strongly disagrees
>		with many of the assertions and conclusions of RFC-1597.
But also agreed upon by a significant part of the Internet community.
It is not an idea from the four of us; we have just written down
some practice that has been in use for quite some time.

I agree with them that global unique addresses are the best solution.
I strongly argue, however, that that might not be possible in situations
as sketched below. I suggest people to read RFC1597 and form their
own opinion.

>	(4) An Internet Draft rebutting RFC-1597 is in preparation by
>		several different members of the Internet community.
What the authors of the contra-document forget to mention is that 
obtaining large amounts of address space is not easy these days.
The rules set out to decide how much address space to assign
(which are set by IETF, IESG and IAB), lead that 90% of the requests
we receive (the RIPE NCC processes all requests for 32C's or more
address space in Europe), are turned down because the assignment efficiency
is too low.

Current guidelines require us to be extremely strict assigning address
space. Some organisations, which have chosen TCP/IP not as a vehicle
to connect tot the Internet, but because of the availability of 
mature implementations, do not want to assign as efficient as the 
guidelines are. With RFC1597, they have two possibilities:
- re-designing their network to comply with current assignment rules;
  this will yield them something that they have indicated they do not
  want (full connectivity), but makes network administration less
  convienent;
- use RFC1597, and be as liberal assigning address space as they want
  to be. Internet communication is still possible using firewall
  constructions such as the original requestor suggested.

Everybody should make this choice on their own. The only choice that
is no longer available is to be liberal with public address space..

>My own quick analysis of RFC-1597 led me to find what I think is a new
>and significant potential security attack arising when a typical
>firewall is used with networks configured as RFC-1597 suggests.  
>I haven't confirmed the existence of the attack yet myself, so I
>deliberately use the word "potential" here.

I would like you to elaborate on this. If a security hole is there,
then many sites are affected since firewalling is common practice
and I think people want to know about it.
I don't see how assigning address X over Y would make any difference
with respect to security.

Geert Jan


-----------[000081][next][prev][last][first]----------------------------------------------------
Date:      4 May 1994 17:40:20 GMT
From:      jpb10@owl.cs.uofs.edu (James Barrett)
To:        comp.protocols.tcp-ip
Subject:   osi vs tcp/ip

  I am trying to contrast OSI (open systems interconnection) with TCP/IP.
However, I am having difficulty understanding OSI addressing and routing
schemes.  I know the basic TCP address such as 134.198.11.10, but I need
an example of an OSI address.

  My professor told me that I should post this question to see if anyone 
would be able to help.

Thanks,

  Jim Barrett

email: jpb10@cs.uofs.edu 

-----------[000082][next][prev][last][first]----------------------------------------------------
Date:      Wed, 4 May 1994 17:50:12 GMT
From:      geertj@ripe.net (Geert Jan de Groot)
To:        comp.protocols.tcp-ip
Subject:   Re: IP Addresses and the Internet?

In <2q5vom$ki@sophia.inria.fr> huitema@mitsou.inria.fr (Christian Huitema) writes:
>As for your initial question on "how many addresses do I need", I suggest you
>play safe and get coordinated addresses for your whole net. The cost is
>minimal and the nightmares you could get otherwise may be quite interesting.
>Just suppose you buy another company or that you decide to establish open
>connectivity with some of your providers. Or even that Internet security
>suddenly increases to the point where you don't need the firewall at all :-)

Note that obtaining official address space is no guarantee that you
will need to renumber your network in the future. In the current classfull
environment, if you would obtain a class B and part of your company
would be sold, that part would probably still need to renumber.

Also, renumbering-while-changing-serviceprovider might become common
practice if the holes-in-CIDR-blocks problem becomes excessive.
Many routers are running on their limits with the size of the current
global routing table. CIDR is working to make these tables smaller,
but many holes will probably bring back the problem. This is why
renumbering might become common practice in the future.

Geert Jan


-----------[000083][next][prev][last][first]----------------------------------------------------
Date:      4 May 1994 19:35:14 GMT
From:      adiwan@siacbbn.com (Arif Diwan (BBN))
To:        comp.protocols.tcp-ip
Subject:   Re: Network Time Synchronization

In article <2q5tfi$epo@news.roadnet.ups.com>,
	wdp@roadnet.ups.com (Bill Paris) writes:
 I'm seeking information on network clock synchronization.

 Our goal is to implement a mechanism for synchronizing the clocks
 of the computers (non-Unix) in our TCP/IP network. We'll use the
 traditional approach: one computer is assigned master time keeper
 status, the others reset their clocks according to some protocol.

 I've read several Internet RFCs, in particular RFC 958 "Network Time
 Protocol (NTP)" by D.L. Mills, and have obtained the software for the
 BSD Time Synchronization Protocol "timed" server. My questions are
 as follows:

 1. We see our choices as being either porting BSD's timed to OS/2 or
 implementing NTP from scratch. Are there other options I'm unaware
 of? Any commercial products?

 2. BSD's timed uses IP's ICMP timestamp messages, but Mills in RFC 956
 "Algorithms for Synchronizing Network Clocks" says that support for
 these messages in the DoD Internet protocol suite is optional, and
 not generally supported. Is this true? Does this imply that timed
 is of limited use?
<<<

Instead of timed use xntp and there are shareware OS/2 client programs
which will keep the clocks synched. Use archie or browse on Compuserve
to get the NTP code for OS/2, Windows, DOS etc.




-- 

				-- Arif Diwan (adiwan@bbn.com)
						617-873-6274

-----------[000084][next][prev][last][first]----------------------------------------------------
Date:      4 May 1994 19:54:59 GMT
From:      adiwan@siacbbn.com (Arif Diwan (BBN))
To:        comp.protocols.tcp-ip
Subject:   Re: SLIP implementation on SUN OS

In article <1994May4.112036.7602@dcs.warwick.ac.uk>,
	kkchin@dcs.warwick.ac.uk (K K Chin) writes:
 can anyone please help to answer some question on the SLIP
 implementation on sun os :

 1. Is SLIP actually implemented ? If yes is it a standard
 component or an option ?
	As far as I know - no. (i.e. upto Sun OS 4.1.3)
	However, there are shareware/freeware versions available
	I use MorningStar Technologies (MST) ppp (which also
	supports slip and various flavors of cslip.

 2. Will SLIP filter out the IP packet that went out the
 serial port, since only one host will be connect to that port?
 Say if a SLIP interface is setup to connect to host A will
 the packet for host B be sent out through this port (and
 leave the SLIP interface at host A to filter out the unwanted
 packet) ?

     Hmmmm. Only packets destined for the other end of the SLIP
	link will be forwarded to the SLIP interface. In MST's
	case, when you fire up the daemon, it sets up the appropriate
	routes, and tears them down when you bring it down. Here
	is my routing table:

	Destination          Gateway              Flags    Refcnt Use        Interface
	   localhost            localhost            UH       0      2205       lo0
	   blowtorch.bbn.com    trajan.bbn.com       UH       1      335        du0
	   default              86t_bcna             UG       1      48266      le0
	   bbn-net              blowtorch.bbn.com    UG       1      4954       du0
	   159.125.0.0          siac                 U        3      44645      le0

	The second routing table entry is automatically entered and removed by
	MST ppp.

	trajan is the local end of the slip link, and blowtorch the remote end. I added
	the bbn-net route, so my slip link acts as a router.

-- 

				-- Arif Diwan (adiwan@bbn.com)
						617-873-6274

-----------[000085][next][prev][last][first]----------------------------------------------------
Date:      4 May 1994 20:02:47 GMT
From:      adiwan@siacbbn.com (Arif Diwan (BBN))
To:        comp.protocols.tcp-ip,comp.dcom.sys.cisco
Subject:   Re: multiple gateways

In article <id.84791.C8D@nmti.com>,
	ihsan@nmti.com (jaleel ihsan) writes:
j.ihsan|> In article <2q66d4$evl@agate.berkeley.edu>
j.ihsan|> sean@oak.his.ucsf.edu writes:

sean|> Situation: multiple LANs interconnected via cisco routers and T1s.

sean|> Problem: most hosts (PCs, Macs, terminal servers) only know about one
sean|> 'gateway' off the local net. If a router dies the hosts don't know
sean|> about another one on the same LAN. From the end user perspective there
sean|> is no redundancy.

sean|> Solutions?

sean|> Run a gateway discovery protocol. Not available on most non unix hosts.

sean|> Configure the hosts with a 0.0.0.0 netmask and let the routers do proxy
sean|> ARP. The hosts will lose connections when a router dies, but can find a
sean|> new gateway by bouncing. Any reasons not to do this?

j.ihsan|> Cant do this if a temporary break in the connection is unacceptable.

sean|>           Any other suggestions?

j.ihsan|>        I am looking for a better solution too !
<<<

We had this problem too. We use two methods to deal with this situation and these
may not be viable for many people. We do router discovery!

 1. Routers RIP default routes and the traffic is either load balanced (if the
	metrics are equal) or a preferred path is used until there is a break.

 2. We wrote an OSPF listener which has nifty algorithm to pick the best route
	or load balance.

The second method is preferable if you use OSPF (we use variable length subnet
masks, and OSPF is essential for routing in such an environment).

-- 

				-- Arif Diwan (adiwan@bbn.com)
						617-873-6274

-----------[000086][next][prev][last][first]----------------------------------------------------
Date:      Wed, 4 May 1994 20:43:06 GMT
From:      c352c03@hawk.depaul.edu (Alex L. Kudlov)
To:        comp.protocols.tcp-ip
Subject:   Need BOOTP Help!

I need info on BOOTP, especially in Novell Netware environment.
What's involved in setting up a BOOTP server,what software/hardware 
is needed, ets.  If there're any helpful textfiles or FAQs on BOOTP, 
I'd like to know where I can get them?

I will appreciate any information.


-----------[000087][next][prev][last][first]----------------------------------------------------
Date:      Wed, 4 May 1994 23:50:45 GMT
From:      carson@tron.bwi.wec.com (Dana Carson)
To:        comp.protocols.tcp-ip
Subject:   Re: Bargains on Comer's 3 Volume TCP/IP set

I got the Comer books before the first monthly mailer arrived.  Maybe
they are having more of a response then they expected and there on back
order?  You might want to check if they think that should have them
already.

-----------[000088][next][prev][last][first]----------------------------------------------------
Date:      5 May 1994 03:10:56 GMT
From:      p4bruel@srv.PacBell.COM (Paul B Ruel)
To:        comp.protocols.tcp-ip
Subject:   Need throughput information and/or pointers

Hello.  I need to determine the typical data throughput rate using ftp in
both ASCII and binary modes for a given medium throughput rate of R.  What
I guess I'm really asking is for the ratio of data to protocol.  The number
may be empirical or the result of some scholarly simulation, but I need it
rather quickly.  Does anyone have this information?  Can you point me to a
readily available source (book, tech report, RFC, &c) if you haven't got
the answer at hand?  Thanks in advance.  E-mail is welcomed: p4bruel@lvas.
srv.pacbell.com.
							Paul Ruel

-----------[000089][next][prev][last][first]----------------------------------------------------
Date:      5 May 1994 04:02:27 GMT
From:      barmar@think.com (Barry Margolin)
To:        comp.protocols.tcp-ip
Subject:   Re: IP Addresses and the Internet?

In article <herbrCp7FnC.JME@netcom.com> herbr@netcom.com (Herb Rosenberg) writes:
>What aboout all of these large organizations that have Class B 
>assignments?  Surely they have a firewall to prevent authorized access 
>into there systems, so why would they need a coordinated IP address?

Many have "one-way" firewalls, which prevent incoming access, but allow
outgoing access.

Also, even with a complete firewall, you still don't want your internal
addresses to duplicate some external addresses, since the firewall has to
be able to reach both internal and external systems.
-- 
Barry Margolin
System Manager, Thinking Machines Corp.

barmar@think.com          {uunet,harvard}!think!barmar

-----------[000090][next][prev][last][first]----------------------------------------------------
Date:      Thu, 5 May 1994 06:37:37 GMT
From:      asim@hood.ee.umn.edu (Asim L. Beg)
To:        comp.protocols.tcp-ip
Subject:   How do I find the port#'s?


Hi,

Does anyone how I can find out the ports which are in use by the
servers/daemons on a certain machine. I have tried netstat but that
only gives me the active sockets.../etc/inetd.conf wont help because
the servers I'm interested in knowing about are not the standard ones.
They are using port numbers above 1023..
By the way I have an account on the host I need to know info about..
Also...suppose I didnt have an account on a given host and I wanted
to know about the port numbers which are being listened.     

Thanks...







--
Asim L. Beg                      The amount of expertise varies in inverse 
albeg@ee.umn.edu                 ratio to the number of statements understood
asim@mermaid.micro.umn.edu       by general public.
                                               GUMMIDGE'S LAW.

-----------[000091][next][prev][last][first]----------------------------------------------------
Date:      5 May 94 13:26:36 MDT
From:      u-mwong%peruvian.cs.utah.edu@cs.utah.edu (Michael Wong)
To:        comp.protocols.tcp-ip,comp.protocols.tcp-ip.ibmpc
Subject:   Looking for TCP/IP Stack Allowing More Than 64 Sockets


   Hi...anybody know of a company that provides more than 64 sockets in
their TCP/IP stack for DOS?  I'm trying to find one that will provide 
128 sockets.  I'm using Novell's TCP/IP stack in LWP for DOS, which allows 
for 64 sockets; but my application for a project requires me to have more 
than that.  I'm NOT looking for a public domain or shareware stack.

   Any leads are greatly appreciated!

-Mike
-- 
Michael Wong 
u-mwong@peruvian.cs.utah.edu
University of Utah

-----------[000092][next][prev][last][first]----------------------------------------------------
Date:      Thu, 05 May 94 14:00:20 EST
From:      stein@gcomm.com
To:        comp.protocols.tcp-ip
Subject:   Re: Bargains on Comer's 3 Volume TCP/IP set


To get Comer's 3-volume set, you might try "telnet books.com" and see if
Books Unlimited has them at a good price.  You might also look there for
W.R.Stevens' "TCP/IP Illustrated Volume 1" which just came out this year
and is widely praised.

===============================================================================
| ... The Galacticomm Demo System - 305.583.7808 - Home of The Major BBS! ... |
===============================================================================


-----------[000093][next][prev][last][first]----------------------------------------------------
Date:      6 May 1994 00:35:57 -0700
From:      phil@lykos.netpart.com (Phil Trubey)
To:        comp.protocols.tcp-ip
Subject:   Re: Encyption of data between nodes across the net

In article <DRW.94Apr24142356@nevanlinna.mit.edu>,
Dale R. Worley <drw@nevanlinna.mit.edu> wrote:
>In a perfect world, the routers that gateway from Office A to the
>outside world and from Office B to the outside world would
>encrypt/decrypt all outgoing packets that were to/from the other
>office.  But I doubt that such a router exists.

From the Morningstar Web Server (http://www.morningstar.com/):

Our Internet connectivity products provide all-important
security features, such as a robust firewall packet filter
(keeps prying eyes off your LAN), link peer authentication
(verify the identity of the host on the other end of the wire),
selective gateway encryption (ensures the privacy of all
communications that traverse the Internet between your LAN and
your branch offices and business partners), and other useful
tools.

If I'm not mistaken, that last feature refers to Morningstar's 
ExpressRouter which does selective TCP level encryption between 
compatible routers.

Disclaimer: I have no connection with Morningstar, I just just 
happen to read their Web server this morning...

______________________________________________________________________

 Phil Trubey                 | 
 NetPartners                 |
                             | Providing independent consulting in the    
 E-mail: phil@netpart.com    |   application of Internet technology        
 Phone:  714-759-1641        |                                             
 Fax:    714-644-0577        |
______________________________________________________________________
-- 
 

-----------[000094][next][prev][last][first]----------------------------------------------------
Date:      5 May 1994 13:12:28 GMT
From:      Per.Weisteen@hda.hydro.com (Per Weisteen)
To:        comp.protocols.tcp-ip
Subject:   Re: How do I find the port#'s?

In message <CpBH63.4rG@news.cis.umn.edu> - asim@hood.ee.umn.edu (Asim L. Beg) w
rites:
>
>Hi,
>
>Does anyone how I can find out the ports which are in use by the
>servers/daemons on a certain machine. I have tried netstat but that
>only gives me the active sockets.../etc/inetd.conf wont help because
>the servers I'm interested in knowing about are not the standard ones.
>They are using port numbers above 1023..
>By the way I have an account on the host I need to know info about..
>Also...suppose I didnt have an account on a given host and I wanted
>to know about the port numbers which are being listened.     
>
>Thanks...
>
>--
>Asim L. Beg                      The amount of expertise varies in inverse 
>albeg@ee.umn.edu                 ratio to the number of statements understood
>asim@mermaid.micro.umn.edu       by general public.
>                                               GUMMIDGE'S LAW.


Most UNIX'es will show you all ports if you use netstat -a instead of just 
netstat.
  
| Per Weisteen		Email: Per.Weisteen@hda.hydro.com   	|
| Norsk Hydro           Phone: +47 2273 8227                	|
| Norway                					|



-----------[000095][next][prev][last][first]----------------------------------------------------
Date:      5 May 1994 14:47:51 GMT
From:      per@erix.ericsson.se (Per Hedeland)
To:        comp.unix.sys5.r3,comp.protocols.tcp-ip
Subject:   SOLVED: Strange behaviour of TCP/IP in ISC 4.0 - BUT...

The problems we were having turned out to be due to a buggy Ethernet
driver ('ne', for the NE2000) - it would frequently lose packets. I
gather that the driver will be fixed in version 4.1.

However:-), we still have problems with inet sockets on ISC - I guess
it's not strictly a TCP/IP issue, but I can't eliminate it by using unix
sockets since ISC doesn't have them...

Anyway, while I wrote in the previous article that sending data locally
(i.e. both connection endpoints on the same ISC host) worked fine, I
later found out that this isn't true - in particular, write()ing ~1k
chunks to the socket works OK, but with increasing write() sizes, the
throughput goes way *down* (if anything it ought to go *up*, of course).

Below are some timings (elapsed time in whole seconds measured by the
receiver) using 'ttcp' to send/receive 1MB of data locally on an ISC
system and on a Sun, and between the two, with various combinations of
'nbuf' write()s of 'buflen' bytes each.

I should add that a couple of replies to my previous article suggested
increasing the "dblocks" streams parameters - of course that didn't help
with a buggy Ethernet driver; unfortunately it doesn't help with this
problem either.

                      ISC->ISC     ISC->Sun     Sun->ISC     Sun->Sun
 nbuf  buflen           secs         secs         secs         secs
====== ======           ====         ====         ====         ====
 65536     16            42           35            9           13 
 32768     32            23           18            5            7 
 16384     64            13           10            4            4 
  8192    128             7            6            3            2 
  4096    256             4            4            2            2 
  2048    512             9            3            3            1 
  1024   1024             4            2            2            0 
   512   2048            14 !          2            2            0 
   256   4096            51 !!         2            2            0 
   128   8192            50 !!         2            2            0 
    64  16384            51 !!         2            2            0 
    32  32768            50 !!         2            2            0 
    16  65536            50 !!         2            2            0 

Again, any help is appreciated...

--Per Hedeland
per@erix.ericsson.se  or
per%erix.ericsson.se@sunic.sunet.se  or
...uunet!erix.ericsson.se!per

-----------[000096][next][prev][last][first]----------------------------------------------------
Date:      Thu, 5 May 1994 15:00:14 GMT
From:      rturner@qms.com (Randy Turner)
To:        comp.protocols.tcp-ip
Subject:   Netware/IP product


I was under the impression that  Netware/IP from Novell was just an NLM that allowed
you to tunnel netware packets through IP to take advantage of existing WAN IP networks.

Is this true? or does it offer more functionality than just a router?

Thanks!

Randy

---


/* Rt */




-----------[000097][next][prev][last][first]----------------------------------------------------
Date:      5 May 1994 18:11:11 GMT
From:      s530577@tfh-berlin.de (Martin_Kliemann)
To:        comp.protocols.tcp-ip
Subject:   MPR IPX/SPX-TCP/IP


Hi,
I'm looking for free (our college has no money) Multi Protocol Router software. We want to connect our Novell-net via a dedicated PC with EtherCard Elite 16 ultra cards (as router).
Thanks for any information

Martin Kliemann
e-Mail: s530577@sun09.tfh-berlin.de



-----------[000098][next][prev][last][first]----------------------------------------------------
Date:      6 May 1994 08:04:37 -0400
From:      johnk@opel.secondsource.com (John Kennedy)
To:        comp.unix.sys5.r3,comp.protocols.tcp-ip
Subject:   Re: SOLVED: Strange behaviour of TCP/IP in ISC 4.0 - BUT...

In article <2qb0uo$jgc@euas20.eua.ericsson.se> per@erix.ericsson.se (Per Hedeland) writes:
>                      ISC->ISC     ISC->Sun     Sun->ISC     Sun->Sun
> nbuf  buflen           secs         secs         secs         secs
>====== ======           ====         ====         ====         ====
> 65536     16            42           35            9           13 
> 32768     32            23           18            5            7 
> 16384     64            13           10            4            4 
>  8192    128             7            6            3            2 
>  4096    256             4            4            2            2 
>  2048    512             9            3            3            1 
>  1024   1024             4            2            2            0 
>   512   2048            14 !          2            2            0 
>   256   4096            51 !!         2            2            0 
>   128   8192            50 !!         2            2            0 
>    64  16384            51 !!         2            2            0 
>    32  32768            50 !!         2            2            0 
>    16  65536            50 !!         2            2            0 
>

Seems that when buflen > MTU, and the IP level has to generate multiple
packets, things go downhill.  Check with netstat -i for the MTU for
your particular device.  (Can that be changed?)  You may be able
to decrease that value and see if the performance drops off at
a different point.

-- 
John Kennedy                     johnk@secondsource.com 
Second Source, Inc.
Annapolis, MD

-----------[000099][next][prev][last][first]----------------------------------------------------
Date:      Fri, 06 May 1994 09:56:36 -0400
From:      opie@panix.com (Matthew S. "Opie" Warren)
To:        comp.protocols.tcp-ip,comp.os.ms-windows.networking.tcp-ip,comp.sys.mac.comm,comp.sys.netware
Subject:   BootP server wanted

Hi.

I have just about given up on Novell's bootp server.  So, I am looking for 
a bootp server that does the following things:

- Works on multiple subnets
- Restricts itself to certain ranges (e.g. assigns only
XXX.YYY.ZZZ.150-253)
- Works on PCs and Macs
- Does _not_ enforce records from previous boots.  i.e. If I boot a machine
  in subnet X, and then boot it in subnet Y, the server should just give
  it a subnet Y IP number, and not quibble about the fact that it already
  has an entry in its bootptab file.  (If I didn't want a new IP number,
  I wouldn't have moved the machine!)
- Is either commercial or shareware/freeware


The operating systems I would want to use, in order of preference, are:

1.  DOS/Windows
2.  Mac OS
3.  Unix (AIX)
4.  NetWare

Do you have any suggestions?


Thanks very much,
Matt

-- 
Matthew S. "Opie" Warren     opie@panix.com     (Mail for PGP key)
"It wouldn't have been anything, even if it were gonna be a thing."

-----------[000100][next][prev][last][first]----------------------------------------------------
Date:      6 May 1994 06:13:30 GMT
From:      bittner@.irf.uni-dortmund.de ()
To:        comp.protocols.tcp-ip,comp.protocols.tcp-ip.ibmpc
Subject:   PD-RPC for DOS or Windows

Hi Folks !

I need a rpc-Version for DOS and/or for Windows in order to build
rpc-clients AND rpc-server. Sun`s PC-NFS doesn't offer the rpc-server-mode.
Are there solutions out there ? Public Domain or commercial ones ?

Christian A. Bittner


---
Christian A. Bittner                     -------------------------------------------  eMail: 					   ---- "It's not a bug - it's a feature" ----
bittner@marius.irf.uni-dortmund.de         ---- A programmer's famous last words  ----
                                           -------------------------------------------

-----------[000101][next][prev][last][first]----------------------------------------------------
Date:      Fri, 06 May 94 12:02:21 EST
From:      stein@gcomm.com
To:        comp.protocols.tcp-ip
Subject:   tip with cslip problems


Regarding Scott Cechovic's question on tip/cslip-2.7 ignoring lost
carrier:

&C1 means the DCD signal goes low when carrier is lost on all modems
I've heard of recently.  Is this an external modem?  I'd intercept the
"tip" end with an LED-RS-232 tester if so, to verify DCD is connected and
working as expected.  If you see DCD go off, and tip doesn't react, it
must be ignoring it.

I have noticed with the SLIP connection of the Trumpet WINSOCK that
"SLIP ENABLED" seems to be the quiescent state -- even when the modem is
disconnected.  Perhaps tip is intentionally ignoring DCD?

-- Bob Stein

===============================================================================
| ... The Galacticomm Demo System - 305.583.7808 - Home of The Major BBS! ... |
===============================================================================


-----------[000102][next][prev][last][first]----------------------------------------------------
Date:      6 May 1994 07:36:15 GMT
From:      etxmesa@eos.ericsson.se (Michael Salmon)
To:        comp.protocols.tcp-ip
Subject:   Re: How do I find the port#'s?

In article <CpBH63.4rG@news.cis.umn.edu>
asim@hood.ee.umn.edu (Asim L. Beg) writes:
|> 
|> Hi,
|> 
|> Does anyone how I can find out the ports which are in use by the
|> servers/daemons on a certain machine. I have tried netstat but that
|> only gives me the active sockets.../etc/inetd.conf wont help because
|> the servers I'm interested in knowing about are not the standard ones.
|> They are using port numbers above 1023..

Have you looked in /etc/services? In any case netstat can give you the
info you are interested in, from the man page:

     -a             Show the state of all sockets; normally sock-
                    ets used by server processes are not shown.

-- 

Michael Salmon

#include	<standard.disclaimer>
#include	<witty.saying>
#include	<fancy.pseudo.graphics>

Ericsson Telecom AB
Stockholm

-----------[000103][next][prev][last][first]----------------------------------------------------
Date:      Fri, 6 May 94 13:25:28 EDT
From:      bpombrio@VNET.IBM.COM
To:        comp.protocols.tcp-ip,comp.protocols.tcp-ip.ibmpc
Subject:   Re: Looking for TCP/IP Stack Allowing More Than 64 Sockets

In article <1994May5.132637.24892@hellgate.utah.edu>,
 on 5 May 94 13:26:36 MDT,
 Michael Wong <u-mwong%peruvian.cs.utah.edu@cs.utah.edu> writes:
>
>   Hi...anybody know of a company that provides more than 64 sockets in
>their TCP/IP stack for DOS?  I'm trying to find one that will provide
>128 sockets.  I'm using Novell's TCP/IP stack in LWP for DOS, which allows
>for 64 sockets; but my application for a project requires me to have more
>than that.  I'm NOT looking for a public domain or shareware stack.
>
>   Any leads are greatly appreciated!
>
>-Mike
>--
>Michael Wong
>u-mwong@peruvian.cs.utah.edu
>University of Utah

Take a look at the IBM offering.  Our sockets are DOS file handles and you
can have 250 of those.

Bob Pombrio                 | 1 daughter 2 daughter 3 little... Nooooo
TCP/IP for DOS Development  | please make this one a boy!  I am so out
bpombrio@vnet.ibm.com       | numbered now :)

"I make no statements or commitments for IBM Corp., nor does IBM make
 statements or commitments for me."

-----------[000104][next][prev][last][first]----------------------------------------------------
Date:      Fri, 6 May 1994 10:40:20 GMT
From:      wdawson@willard.atl.ga.us (Willard Dawson)
To:        comp.unix.sys5.r3,comp.protocols.tcp-ip
Subject:   Re: SOLVED: Strange behaviour of TCP/IP in ISC 4.0 - BUT...

per@erix.ericsson.se (Per Hedeland) writes:

>The problems we were having turned out to be due to a buggy Ethernet
>driver ('ne', for the NE2000) - it would frequently lose packets. I
>gather that the driver will be fixed in version 4.1.

A friend tells me that the NE2000 driver has always been buggy, and also
told me to avoid NE2000 cards like the plague.  I suppose that "just
wait 'til next year" clause still applies...

-----------[000105][next][prev][last][first]----------------------------------------------------
Date:      Fri, 6 May 1994 10:59:55 GMT
From:      tke@purple.srl.utu.fi (Timo Eronen)
To:        comp.protocols.tcp-ip
Subject:   Re: pcnfsd

In <Cp1z9u.JKL@csn.org>, davidk@csn.org (David Kirkpatrick) writes:
>
>Does anyone know where to get source code for pcnfsd?  I need to connect
>a PC running DOS/Windows to a Stratus running their version of UNIX
>(called FTX).  Problem is, FTX doesn't come with pcnfsd, so I guess I
>need to try to compile it myself.  I'm running PC/TCP version 2.3 on the
>Windows machine, and already have tcp/ip running between the two 
>machines....

You don't need to run PCNFSD on Stratus. Instead, you can configure your
PC/TCP to use NIS host at some other machine.

Cheers, Timo

--------------------------------------------------------
Timo Eronen, tke@utu.fi
Space Research Laboratory, University of Turku, Finland.


-----------[000106][next][prev][last][first]----------------------------------------------------
Date:      6 May 1994 11:19:40 +0100
From:      pmiles@tdc.dircon.co.uk (Peter Miles)
To:        comp.unix.programmer,comp.protocols.tcp-ip
Subject:   IP address/sitename from network login

Can anyone suggest how, using C on a UNIX system (System V with sockets) 
I can get details of the originating IP address (and hostname) of a 
network login, AFTER the user has logged in using telnet or rlogin.

Basically, I need a C program  which when run, returns the IP address of 
the user's originating system.

I know you can get this information with TCP wrapper, but that only works
BEFORE the user has logged in -- I need this information after they've 
logged in.

If anyone has any ideas (or example code), I'd be extremely grateful.
Please email replies, and I'll post a summary.

		-- Pete

-- 
Pete Miles			pmiles@dircon.co.uk
				...pipex!dircon!pmiles

-----------[000107][next][prev][last][first]----------------------------------------------------
Date:      6 May 1994 13:10:48 GMT
From:      carlson@Xylogics.COM (James Carlson)
To:        comp.unix.sys5.r3,comp.protocols.tcp-ip
Subject:   Re: SOLVED: Strange behaviour of TCP/IP in ISC 4.0 - BUT...

In article <2qdbol$ip6@opel.secondsource.com>, johnk@opel.secondsource.com (John Kennedy) writes:
|> Seems that when buflen > MTU, and the IP level has to generate multiple
|> packets, things go downhill.  Check with netstat -i for the MTU for
|> your particular device.  (Can that be changed?)  You may be able
|> to decrease that value and see if the performance drops off at
|> a different point.

You must be joking!  They're fragmenting at the IP level when going
through a TCP socket?!  That sounds like a broken implementation to me!

--
James Carlson <carlson@xylogics.com>            Tel:  +1 617 272 8140
Annex Software Support / Xylogics, Inc.               +1 800 225 3317
53 Third Avenue / Burlington MA  01803-4491     Fax:  +1 617 272 2618

-----------[000108][next][prev][last][first]----------------------------------------------------
Date:      Fri, 6 May 1994 13:34:02 GMT
From:      grimes@netcom.com (George Grimes)
To:        comp.unix.programmer,comp.protocols.tcp-ip
Subject:   Re: IP address/sitename from network login

In article <2qd5js$etj@tdc.dircon.co.uk> pmiles@tdc.dircon.co.uk (Peter Miles) writes:
>Can anyone suggest how, using C on a UNIX system (System V with sockets) 
>I can get details of the originating IP address (and hostname) of a 
>network login, AFTER the user has logged in using telnet or rlogin.
>
>Basically, I need a C program  which when run, returns the IP address of 
>the user's originating system.
>
>I know you can get this information with TCP wrapper, but that only works
>BEFORE the user has logged in -- I need this information after they've 
>logged in.
>
>If anyone has any ideas (or example code), I'd be extremely grateful.
>Please email replies, and I'll post a summary.
>
>		-- Pete
I wanted to do something like this to automatically set my DISPLAY environment
variable when I'm running X.  I used the ttyname() call to get the current tty
that I'm connected to then read the utmp file looking for the matching entry.
I do automatically in my .login file.

See "man ttyname" and "man utmp" for more information.  Send me email if  you
have specific questions.

George



-- 
---------------------------+-----------------------------------------+
George A. Grimes           | You may try this UNREGISTERED .sig file |
LTX Corporation            | free for 30 days.  After 30 days, you   |
801 E. Plano Parkway, #160 | must either send in your registration   |
Plano, TX 75074            | fee or remove this .sig file from all   |
(214)578-1006              | your communications.  Registered users  |
grimes@netcom.com          | of this .sig file are entitled to a     |
                           | free copy of the next release.          |
---------------------------+-----------------------------------------+


-----------[000109][next][prev][last][first]----------------------------------------------------
Date:      Fri, 6 May 1994 14:50:51 GMT
From:      slemieux@etu.gel.ulaval.ca (Steve Lemieux)
To:        comp.protocols.tcp-ip
Subject:   Broadcast with different ip source

I have a big problem!

I'm using an HP workstation with 2 lan cards as an ethernet gateway.
On one side, we have a router who doesn't know about the other
network. I'd like to be able to broadcast from the workstation
with ip address from the other side of the gateway to show
(some) machines on this network to the router. We can't use RIP...

We tried Traceroute but we can't specify a source address different
from IP card on the workstation.

So, it is possible to do that?? Maybe writing directly to the lan
driver??

(We prefer a solution using some kind of broadcast send by the workstation
with fake source IP address)

-----------------------------------------------------------------
Steve Lemieux                     BIX       : slemieux
911 development team              Internet  : slemieux@qc.bell.ca
Bell Canada                       Phone     : (418)682-4519
HP-9000 715/817/827/897           Fax       : (418)688-9977
-----------------------------------------------------------------


-----------[000110][next][prev][last][first]----------------------------------------------------
Date:      6 May 1994 17:02:39 GMT
From:      barmar@think.com (Barry Margolin)
To:        comp.protocols.tcp-ip
Subject:   Re: Telnet Echo Option

In article <roberto.2.00108F4B@jsb.co.uk> roberto@jsb.co.uk (Robert Osborne) writes:
>Can anybody help me with a problem I am having with the Telnet echo option. If 
>I turn off all echoing, local and remote, everything appears to be fine upon 
>the initial login. No characters are echoed at all, as I expect to happen. 
>
>However, once the login process has been completed, and I am at the usual 
>command line prompt, whatever is typed by the user is echoed back. I have not 
>received any command telling me that remote echoing has been turned back on.

Do you use a shell that implements Emacs- or Vi-style command line editing?
If so, it may be doing its own echoing, rather that depending on the tty
driver to do automatic echoing.  The TELNET Echo option is only manipulated
by the telnet daemon when the tty mode changes.

One way to test this would be to see whether you still get echoing when you
go into a simple (non-full-screen) interactive application, such as "ed".
If the shell is doing its own echoing, you shouldn't see any echoing when
running one of these programs.
-- 
Barry Margolin
System Manager, Thinking Machines Corp.

barmar@think.com          {uunet,harvard}!think!barmar

-----------[000111][next][prev][last][first]----------------------------------------------------
Date:      6 May 1994 17:43:09 GMT
From:      hch@hybrid.com (Howard Hart)
To:        comp.protocols.tcp-ip,comp.protocols.tcp-ip.domains
Subject:   Re: DNS and variable length subnets

In article <2prvgl$l9o@babs.nsr.hp.com>, cricket@nsr.hp.com (Cricket Liu) writes:
|> Russell Sutherland (russ@madhaus.uucp) wrote:
|> 
|> : Do these four bits of contiguous ip address space have to served by one (and
|> : not four) name server(s)?
|> 
|> You could also delegate at the last octet, though that's messy.  In other
|> words, you could delegate d.c.b.a.in-addr.arpa from the c.b.a.in-addr.arpa
|> name server.  That implies, however, that you've got as many subdomains as
|> you do hosts whose IP addresses begin with a.b.c.  Yech.
|> 
|> --
|> cricket
|> 
|> cricket@hp.com / Hewlett-Packard Professional Services / Englewood, CO

So if I understand this, it isn't possible to have two or more subdomains
serving one Class C, subnetted or not? I have what I thought was a 
working model of this, two primaries, bcw.hybrid.com and dca.hybrid.com
on one class C, with dca.hybrid.com being primary for the IN-ADDR
record they shared. I am seeing problems though in that the 
bcw.hybrid.com nameserver refuses to be authoritative, even for
itself (goes all the way to BARRNET for name resolution). 

Am I doing something that will never really work? Thanks in advance.

---
Howard Hart                                UUCP:ames!hybrid.com!hch
Network Administrator                      INTERNET: hch@hybrid.com
Hybrid Networks, Inc.                      PHONE: (408) 725-3257
Cupertino, CA

-----------[000112][next][prev][last][first]----------------------------------------------------
Date:      Fri, 6 May 1994 17:53:01 GMT
From:      root@news.uab.es (Gloria Hernandez. proyecto informatica)
To:        comp.protocols.tcp-ip
Subject:   Programming with RPC. (PROBLEM)

Hello: 


I am trying to understand the rpc Protocol Compiler  "rpcgen".
But, I have a problem when I run an example application   (rprintmsg)
The server program is running perfectly, but when i try to run the
client program it displays the error message:


		"Illegal instruction"


when it try to execute the instruction
   
  		result = printmessage_1(cl,&message);

(rprintmsg is the remote procedure)


I have read about "RPC status code" but I dont Know what is the reason.


Please , Can you help me?


I wait for your answers:


			Thank you.

					Gloria Hernandez 
					Universidad Autonoma de Barcelona
					Spain
					EMAIL   gloria@morgana.uab.es


=========================================================================================








-----------[000113][next][prev][last][first]----------------------------------------------------
Date:      Fri, 6 May 1994 19:12:28 GMT
From:      blanchet@fsg.ulaval.ca (Marc Blanchet)
To:        comp.protocols.tcp-ip
Subject:   DHCP server software

Hi,
	does anyone know where I can find a dhcp server for Unix?
(public domain).  I know Sun's SolarNet, but would prefer public
domain.  I tried with archie and nothing appears.


Regards, Marc.

---

------------------------------------------------------------------
Marc Blanchet                  | telephone:   418-656-3559
Coordonnateur a l'informatique | telecopieur: 418-656-3170
Fac. Sciences et Genie         |
Pouliot 1100                   |
Universite Laval               | Internet: Marc.Blanchet@fsg.ulaval.ca
Quebec, Quebec, Canada         | 
G1K 7P4                        |
------------------------------------------------------------------


-----------[000114][next][prev][last][first]----------------------------------------------------
Date:      Fri, 6 May 1994 19:15:21 GMT
From:      dave@odyssey.ucc.ie
To:        comp.protocols.tcp-ip
Subject:   REQUEST: Packet Driver for a DEC Etherworks card.

Can anyone point me to a packet driver (pktint or something ?) for
a DEC etherworks 3 (turbo) card.

thanks
dave
(dave@odyssey.ucc.ie)

-----------[000115][next][prev][last][first]----------------------------------------------------
Date:      Fri, 6 May 1994 20:55:51 GMT
From:      vjs@calcite.rhyolite.com (Vernon Schryver)
To:        comp.unix.sys5.r3,comp.protocols.tcp-ip
Subject:   Re: SOLVED: Strange behaviour of TCP/IP in ISC 4.0 - BUT...

In article <2qdfko$6sl@newhub.xylogics.com> carlson@xylogics.com writes:
>In article <2qdbol$ip6@opel.secondsource.com>, johnk@opel.secondsource.com (John Kennedy) writes:
>|> Seems that when buflen > MTU, and the IP level has to generate multiple
>|> packets, things go downhill.  Check with netstat -i for the MTU for
>|> your particular device.  (Can that be changed?)  You may be able
>|> to decrease that value and see if the performance drops off at
>|> a different point.
>
>You must be joking!  They're fragmenting at the IP level when going
>through a TCP socket?!  That sounds like a broken implementation to me!

Without known anything about the context or whether something or other
is broken, please note that IP fragmenting instead of TCP segmenting
has been used by some vendors to go faster.  It is sometimes faster to
IP fragment.  For example, at one time a network adapter box for a super
computer claimed the fastest TCP/IP performance.  One of their tricks
was to tell the super computer to use a TCP MTU of 32K (I think) and to
IP fragment and reassemble in the "low speed channel to FDDI adapter."


Vernon Schryver    vjs@rhyolite.com

-----------[000116][next][prev][last][first]----------------------------------------------------
Date:      7 May 94 02:01:33
From:      billw@glare.cisco.com (William )
To:        comp.protocols.tcp-ip
Subject:   Re: Telnet Echo Option

    Can anybody help me with a problem I am having with the Telnet
    echo option. If I turn off all echoing, local and remote,
    everything appears to be fine upon the initial login. No
    characters are echoed at all, as I expect to happen.

    However, once the login process has been completed, and I am at
    the usual command line prompt, whatever is typed by the user is
    echoed back. I have not received any command telling me that
    remote echoing has been turned back on.

Ok.  I think you are confused as to how the "echo option" works.  It
isn't so much that it tells you who is echoing data, it tells you who
is RESPONSIBLE for echoing data.

Server	Will Echo	"remote echoing", server echos characters in tty
			driver, parser, or application.  By NOT echoing
			characters, "echo off" is implemented.  But when
			a server says "will echo", there is no real hint
			as to whether it will echo characters, or whether
			it thinks all echoing should be off.
	Won't echo	Client should echo characters to user.
			local echoing -> "I'm not going to echo anything,
			so you better do it!"
Client	Will Echo	Bogus.  Would imply that the client would echo
			data stream characters back to host.
	Won't echo	default (don't be bogus!)

This confusing state of affairs arises from trying to have a symetric
option negotiation protocol for an option that is inherently unsymetric.

BillW
cisco

-----------[000117][next][prev][last][first]----------------------------------------------------
Date:      Fri, 6 May 1994 21:04:09 GMT
From:      alan@ernest.itg.ti.com (Alan Edmonds)
To:        comp.unix.sys5.r3,comp.protocols.tcp-ip
Subject:   Re: SOLVED: Strange behaviour of TCP/IP in ISC 4.0 - BUT...

In article <2qdbol$ip6@opel.secondsource.com>,
John Kennedy <johnk@secondsource.COM> wrote:
>Seems that when buflen > MTU, and the IP level has to generate multiple
>packets, things go downhill.  Check with netstat -i for the MTU for
>your particular device.  (Can that be changed?)  You may be able
>to decrease that value and see if the performance drops off at
>a different point.

I don't think there is a way to decrease the MTU (besides getting into
the driver source and hacking away).

-- 
Alan Edmonds, KB5ZUY                             Texas Instruments, Inc.
I don't speak for TI; TI doesn't speak for me    M/S 8515
Work phone: +1-214-575-6427                      6620 Chase Oaks Blvd.
Email: edmonds@lobby.ti.com                      Plano, TX, USA  75023

-----------[000118][next][prev][last][first]----------------------------------------------------
Date:      Fri, 6 May 1994 21:20:23 GMT
From:      rad@tyrell.net (Bob Daniel)
To:        comp.protocols.tcp-ip
Subject:   TCP/IP and X.25 over VSAT Sattelite

We are doing TCP/IP over X.25 but would like to avoid telnet since it
sends one packet per character (plus an ack packet if not local).  We
are looking for alternative ways to do message based terminal emulation
over X.25 between UNIX hosts.

How does the Internet deal with the packet-per-character issue with telnet?
Are there more efficient ways to do telnet?
Are there X.25 terminal emulation programs for UNIX?

Bob Daniel
rad@tyrell.net
-- 

bob.daniel
rad@tyrell.net

-----------[000119][next][prev][last][first]----------------------------------------------------
Date:      Fri, 6 May 94 21:24:04 GMT
From:      mak@nexor.co.uk (Martijn Koster)
To:        comp.protocols.tcp-ip
Subject:   Re: Looking for RFC distrib. site

In article <1994Apr29.092607.3865@datcon.co.uk>, pgw@datcon.co.uk (Peter Waters) writes:
|> Mark C. Stout (mcstout@netcom.com) wrote:
 
|> You can obtain RFC's from nic.ddn.mil using ftp in the directory /rfc
|> 
|> RFC 1600 contains a list of all rfc's.

If you are a WWW user you may also want to try the Searcheable RFC Index
on <http://web.nexor.co.uk/rfc-index/rfc-index-search-form.html>. You
can search titles and descriptions, and the results refer you to 
relevant on-line RFC's at a number of sites.

-- Martijn
__________
Internet: m.koster@nexor.co.uk
X-400: C=GB; A= ; P=Nexor; O=Nexor; S=koster; I=M
X-500: c=GB@o=NEXOR Ltd@cn=Martijn Koster
WWW: http://web.nexor.co.uk/mak/mak.html

-----------[000120][next][prev][last][first]----------------------------------------------------
Date:      Fri, 6 May 1994 22:02:14 GMT
From:      dcm0383@mcdata.com (Dick C. Masyga)
To:        comp.protocols.tcp-ip
Subject:   Re: RS6000 & TOKEN RING

Olivier MARTINERIE (martiner@int-evry.fr) wrote:
: Hi! 
: Here's the problem of a friend:
: * We have 2 rings linked by bridges.
: * A machine connected on ring 2 try to link a risc 6000 on ring 1.
: * This machine send a frame with correct TCP/IP addresses.
: * But the RS6000 does not answer that frame.
: * The same frame, sent by another machine on ring 2, is recognised by the 
: rs6000.
: (machine working on ring 1: IBM Controller
: machine not working: Mac Data controller)
: ANY IDEAS ???
: "
: Thanks for answering, here or in my email !
: Olivier MARTINERIE                                      Amiga 4000-30
: Phone: +33 1 60 76 64 69		          	Amiga 1000	
: Fax: +33 1 46 32 33 68
: IRC Nickname: KAZIM				National Institute of
: E-Mail: martiner@galaxie.int-evry.fr		Telecoms (France)


I have had an opportunity to see some of the traces regarding this problem
and it is very clear that the problem lies in the bridge.  The bridge is 
passing the first ARP sent by the McDATA controller, even though this ARP 
does not have the all routes broadcast bit set, or routing information.
The ARP is seen by the RS6000 as though the sender is on the same ring and
replies as it should without any routing.  This is where the bridge fails. 
It does not send this response back to the original ring.  Since the McDATA
does not get a reply to the first ARP, it makes an attempt to broadcast the 
ARP to all rings.  The bridge once again sends this ARP (with broadcast bit
set and routing information) to the next ring.  This time the RS6000 does not
respond, presumably because it has all ready replied to an ARP from the McDATA.

The McDATA is following RFC #1042 for sending ARPs on IEEE 802 networks.  

In this case, the bridges should either pass all traffic (even if there is no 
routing), or should not pass any traffic unless specifically broadcasted.  You
cannot have one side pass all traffic, but have the other direction filtered
to only send broadcast or routed messages.  If the problem is not in the bridge
configuration, it should be reported to the bridge vendor as a bug.

If the bridge would pass all traffic, the first ARP would make it back to 
the McDATA and the connection would continue.  If the bridge would never pass
traffic unless routed, the first ARP would die, and the next ARP (routed)
would make it to the RS600 and the reply would make it back to the McDATA. 
-----------------------------------------------------------------
	Dick Masyga               | These are my opinions 
	dcm0383@mcdata.com        | Standard disclaimers apply
-----------------------------------------------------------------

-----------[000121][next][prev][last][first]----------------------------------------------------
Date:      Fri, 6 May 1994 23:29:39 GMT
From:      donp@novell.com (don provan)
To:        comp.protocols.tcp-ip
Subject:   Re: Telnet Echo Option

In article <roberto.2.00108F4B@jsb.co.uk> roberto@jsb.co.uk (Robert Osborne) writes:
>Can anybody help me with a problem I am having with the Telnet echo option. If 
>I turn off all echoing, local and remote, everything appears to be fine upon 
>the initial login. No characters are echoed at all, as I expect to happen. 
>
>However, once the login process has been completed, and I am at the usual 
>command line prompt, whatever is typed by the user is echoed back. I have not 
>received any command telling me that remote echoing has been turned back on.

The manually disabled echo is probably being overridden by the OS when
it disables the echo to protect your password.  Once the OS has the
password, my guess is it "reenables" echoing, forgetting that echoing
was initially disabled.

Or, to put it another way, the OS believes that it can change echoing
for its own purposes and does not feel obligated to restore any
previous setting. "That's not a bug, it's a misfeature."

					don provan
					donp@novell.com

-----------[000122][next][prev][last][first]----------------------------------------------------
Date:      Sat, 7 May 1994 01:21:21 GMT
From:      quang@pairgain.com (Quang Bui)
To:        comp.protocols.tcp-ip
Subject:   routed QUESTION

Hello every one,

I have problem getting pass the gateway to get out to the Internet.
Here is the network configuration. On my main subnet which connects to
the Internet I have no problem getting to the Internet from any node,
but from any subnet which connect to the main net, I cannot get out
to the Internet at all, to get around this I have to rlogin to
the main net. Is this a problem with routing, DNS, proxy server, etc...

Thank you in advance


Quang Bui
Pairgain Technologies


-----------[000123][next][prev][last][first]----------------------------------------------------
Date:      Sat, 7 May 1994 01:44:54 GMT
From:      byoder@netcom.com (Brian K. Yoder)
To:        comp.protocols.tcp-ip
Subject:   WANTED: NDIS/ODI Shim

Can anyone direct me to where i can get an ODI driver to live on top
of another NDIS driver?

--Brian
-- 

+------------------+-----------------------------------------------------------+
| Brian K. Yoder   | "The children who know how to think for themselves, spoil |
| byoder@netcom.com|  the harmony of the collective society that is coming,    |
| US Networx, Inc. |  where everyone (would be) interdependent" --John Dewey   |
+------------------+-----------------------------------------------------------+

-----------[000124][next][prev][last][first]----------------------------------------------------
Date:      7 May 1994 13:28:15 -0700
From:      schmidt@tango.ics.uci.edu (Douglas C. Schmidt)
To:        comp.unix.solaris,comp.protocols.tcp-ip
Subject:   FTP programming APIs

Hi,

	I seem to recall there is a publically available library
containing C routines that provide an API for accessing ftp services
within a program (rather that using the ftp program manually).  Could
someone please point me in the right direction for obtaining these
routines?

	Thanks,
	
		Doug
-- 
His life was gentle, and the elements so            | Douglas C. Schmidt
Mixed in him that nature might stand up             | schmidt@ics.uci.edu
And say to all the world: "This was a man."         | ucivax!schmidt
   -- In loving memory of Terry Williams (1971-1991)| (714) 856-4105

-----------[000125][next][prev][last][first]----------------------------------------------------
Date:      7 May 1994 12:29:52 -0500
From:      electro@matt.ksu.ksu.edu (Eric L Patterson)
To:        comp.protocols.tcp-ip
Subject:   PCROUTE and SLIP

Here's the scenerio:
KSU has a slip server running.
I can connect to it with one machine, but I'd like to be able to use up to
4 machines with the line.  I've tried pc-route, but I can't seem to get 
it to work.  
I've connected the modem and other pc via slip to the router, but when I try to
ping anything, i get ICMP transmit errors.  Also, I suspect that
pcroute hangs up, but I can't tell if it has or not.
Anybody have any suggestions?
-- 
Eric Patterson -- electro@ksu.ksu.edu -- N0SJW

-----------[000126][next][prev][last][first]----------------------------------------------------
Date:      7 May 1994 21:11:41 -0700
From:      aevans@kaiwan.com (Alan B. Evans)
To:        comp.protocols.tcp-ip
Subject:   Re: Packet Sniffers...

System Administrator (root@athens.newport.org) wrote:
: Are there any generic UNIX packet sniffers out there (or specifically, for
: Linux)? I am in need of one for troubleshooting purposes on our LAN...
 
: If you intend to reply via e-mail, please use "root@athens.uucp.netcom.com" 
: instead of "root@newport.org".
 
: Thanks, Mike.
: -- 
: ____________________________________________________________________________
: System Administrator                                        root@newport.org
: Network Operations                                        The Newport School 

There is a package called tcpdump out on sunsite (source and binary form)
that does exactly what you are asking. The binaries are compiled to run
under linux all ready.

Alan

-- 
You can reach me at @ :
Internet : aevans@kaiwan.com
ICBM     :   33 39'     North   Cruise  :   33 39' 37"  North
         :  117 59'     West            :  117 59' 54"  West

-----------[000127][next][prev][last][first]----------------------------------------------------
Date:      7 May 1994 10:11:52 GMT
From:      jelson@ren.psy.jhu.edu (Jeremy Elson)
To:        comp.protocols.tcp-ip
Subject:   Re: Bargains on Comer's 3 Volume TCP/IP set

Thayne Forbes <thayne@xmission.com> wrote:
>:> : The Library of Computer and Information Sciences (LCIS), is currently
>:> : having a promo with Comer's 3 books.  You can save some bucks, since the
>:> : Publisher's prices on these is $164!  This offer is open to any
>:> : non-members (wish *I* could).  The scoop is this:
 
>:> The only problem is that I joined in March (maybe Feb) and I'm still waiting
>:> for two of the Volumes
>
>And I am still waiting for all three.  I have, on the other hand, got
>three of the monthly mailers, and had to refuse the three monthly
>selections.  (I am beginning to have reservations).

I'm in exactly the same position - I sent my little form in, they billed
me for the five bucks (plus 8 bucks shipping), and I still haven't seen any
books.  I sent the check to them at least 3-4 months ago.  They did send
me a notice at the beginning of March saying that the books were backordered
and would be available on March 28, but I haven't heard anything since then
(except, as you say, for getting the monthly ads).

Does anyone have a phone number for LCIS?  I'd really like to find out where
my books are.  I, too, am beginning to have reservations.


-----------[000128][next][prev][last][first]----------------------------------------------------
Date:      8 May 1994 02:39:53 -0700
From:      gorgon@crl.com (Zach Copley)
To:        comp.sys.next.sysadmin,comp.sys.next.misc,comp.sys.mac.comm,comp.sys.mac.system,comp.protocols.tcp-ip,ca.unix
Subject:   Setting up a home LAN with NEXSTEP/Unix

Setting up a home LAN with NEXSTEP/Unix
 
I'm trying to setup a local area network in my house using my
NEXTSTEP machine.  I want to have the NEXTSTEP machine be a UNIX host
machine, and let all of the other computers in the house telnet in
over ethernet to use shell accounts.
 
I have a Mac II and a Mac Quadra 700, both on ethernet, and they both
have MacTCP 2.0.4.  I also have a DOS machine on ethernet, but I
don't care about that one as much.  The ethernet network is a simple
daisy chain, terminated at both ends.
 
MacTCP wants an IP address for its machine, a gateway address for
routing, whether to obtain the address manually, dynamically, or from
a server, and a subnet mask (net, subnet and node).  It also wants
domain name servers, but that doesn't seem to be manditory.  I
haven't been able to figure out the right numbers to put in.  The IP
address for my NEXTSTEP machine is 192.187.157.37.  I also run SLIP
on it to do e-mail.
 
All that I want to do is run NCSA Telnet and FTP so that my Macs can
access my NeXT machine.  I can't get either application to work.  The
NeXT machine wont respond.  I don't want to run something like Novell
(overkill for me).  I want to connect all the machines with TCP/IP
and make the NeXT machine the central machine.
 
Is this something very complicated to do?  Can anyone help me? 
 
Please?
 
Thanks, 
 
Zach






-----------[000129][next][prev][last][first]----------------------------------------------------
Date:      7 May 1994 15:32:17 -0000
From:      root@athens.newport.org (System Administrator)
To:        comp.protocols.tcp-ip
Subject:   Packet Sniffers...

Are there any generic UNIX packet sniffers out there (or specifically, for
Linux)? I am in need of one for troubleshooting purposes on our LAN...

If you intend to reply via e-mail, please use "root@athens.uucp.netcom.com" 
instead of "root@newport.org".

Thanks, Mike.
-- 
____________________________________________________________________________
System Administrator                                        root@newport.org
Network Operations                                        The Newport School 

-----------[000130][next][prev][last][first]----------------------------------------------------
Date:      Sat, 7 May 94 20:46:49 GMT
From:      ftlofaro@unlv.edu (Frank Lofaro)
To:        comp.protocols.tcp-ip
Subject:   Re: TCP/IP and X.25 over VSAT Sattelite

In article <1994May6.212023.9340@tyrell.net> rad@tyrell.net (Bob Daniel) writes:
>We are doing TCP/IP over X.25 but would like to avoid telnet since it
>sends one packet per character (plus an ack packet if not local).  We
>are looking for alternative ways to do message based terminal emulation
>over X.25 between UNIX hosts.
>
>How does the Internet deal with the packet-per-character issue with telnet?
>Are there more efficient ways to do telnet?
>Are there X.25 terminal emulation programs for UNIX?
>
>Bob Daniel
>rad@tyrell.net
>-- 
>
>bob.daniel
>rad@tyrell.net

Use telnet in line mode. It uses very little bandwidth.


-----------[000131][next][prev][last][first]----------------------------------------------------
Date:      Sun, 8 May 94 03:10:01 CST
From:      <c482165@monad.missouri.edu>
To:        comp.protocols.tcp-ip,comp.protocols.tcp-ip.ibmpc
Subject:   packet driver protocols

I am currently translating a turbo pascal interface to the Crynrware packet
drivers into a C++ class.  Once I'm done, I want to write my own network
software, such as FTP, Telnet, and possibly a WinSock DLL that uses the
Crynrware drivers as an interface.

Once I have the packet driver interface done, what do I do with it?  I need 
to know about tcp-ip protocols:  bootP, ftp, telnet and the like.

I have the book "Internetworking with TCP/IP" by Douglas E. Comer but it's 
not very helpful from a programming standpoint.  Is there any information 
out there on-line that I can get?  

thanks!

- Matt

-----------[000132][next][prev][last][first]----------------------------------------------------
Date:      8 May 1994 00:40:31 GMT
From:      kwia4000@bronto.zrz.TU-Berlin.DE (Manfred Kwiatkowski)
To:        comp.protocols.tcp-ip
Subject:   Re: PCROUTE and SLIP

In article <2qgj6g$7pi@matt.ksu.ksu.edu>, electro@matt.ksu.ksu.edu (Eric L Patterson) writes:
> Here's the scenerio:
> KSU has a slip server running.
> I can connect to it with one machine, but I'd like to be able to use up to
> 4 machines with the line.  I've tried pc-route, but I can't seem to get 
> it to work.  
> I've connected the modem and other pc via slip to the router, but when I try to
> ping anything, i get ICMP transmit errors.  Also, I suspect that
> pcroute hangs up, but I can't tell if it has or not.
> Anybody have any suggestions?

Yeah. No way! I bet this server hands out a SINGLE IP-address for home fun.
To connect more than one device you need as many addresses and real routing.
You will have to convince the KSU people managing the server...

-- 
Manfred Kwiatkowski         kwiatkowski@zrz.tu-berlin.de

-----------[000133][next][prev][last][first]----------------------------------------------------
Date:      8 May 1994 13:35:26 GMT
From:      csmith@blackplague.gmu.edu (Christian Smith)
To:        comp.sys.next.sysadmin,comp.sys.next.misc,comp.sys.mac.comm,comp.sys.mac.system,comp.protocols.tcp-ip,ca.unix
Subject:   Re: Setting up a home LAN with NEXSTEP/Unix

Zach Copley (gorgon@crl.com) wrote:
: Setting up a home LAN with NEXSTEP/Unix
:  
: I'm trying to setup a local area network in my house using my
: NEXTSTEP machine.  I want to have the NEXTSTEP machine be a UNIX host
: machine, and let all of the other computers in the house telnet in
: over ethernet to use shell accounts.
:  
Ok, here is what you should do.  Since the NeXT is using Slip to get to the
outside world, it has TWO network interfaces, call them slip0 and en0.  Each
of these should have its own ip address.  The slip0 interface will get its
ip from the slip server you connect to.  The en0 interface will get its ip
address from /etc/hostconfig.  Now, you have two choices

1) set up an internal ethernet network with no access to the outside world.
2) set up an internal ethernet network with access to the outside world.

Choice one is easy, choice two involves setting routed on the next, and getting
you slip prrovider to set you up with a subdomaine, which they might or might
not be willing to do.  I can't help you out to much with this one.

As for choice one, do it this way....

Using /etc/hostconfig, set up as

HOSTNAME=somename
INETADDR=xxx.xxx.xxx.1
ROUTER=same as above
IPNETMASK=255.255.255.0
IPBROADCAST=xxx.xxx.xxx.255
NETMASTER=-YES-


Now, you have a router and netmask to give the mac, ie the same as the above.
Ip address for the other machines in you home ethernet should be of the form
xxx.xxx.xxx.n where 1<n<255

You can also set the next up to run bind and named and then have nameservice 
for your home network, but you probably don't need this.

This should all work.  Since your home network won't have any connection to
the outside world, it doesn't matter what the ip network you chose is.  
If you want to try to get the home network connected to the outside world,
talk to your slip provider, and get copies of the Nutshell books on
routing, and named/bind as you will need them.  Frankly, I doubt it is worth
the trouble.  

Amanda (or others) if I goofed on here please let me know, this is coming off
the top of my head...

Chris






-----------[000134][next][prev][last][first]----------------------------------------------------
Date:      Sun, 08 May 1994 13:55:44 GMT
From:      mark@cyantic.com (Mark T. Dornfeld)
To:        comp.protocols.tcp-ip
Subject:   Subnet masks NOT on octet boundaries

I currently have the task of reconfiguration of three pairs of Cisco
routers with each pair connecting a local and remote LAN.  TCP/IP and
Ethertalk are the current protocols being routed.

In order to conserve the Class C IP numbers that have been assigned, I
would like to use just one Class C address for the WAN (the Cisco serial
ports) and set up a subnet mask that uses part of the fourth octet. 
I'll use one of the most common examples in the TCP/IP documentation and
select two bits from the last octet to be part of the subnet mask. So the
mask would look like this: 11111111.11111111.11111111.11xxxxxx if
I understand the method correctly. So in decimal, the mask is
255.255.255.192, correct?

The assigned Class C network number (fictitious here) that I am working
with is 192.139.228.  If the mask is correct above, then how are the subnet
number and the node number expressed decimal for the purpose of
configuration?

I assume that I have the range of 1-191 for hosts and 192-223 for network
numbers in the last octet (Numbers over 223 are not supposed to be used for
network addresses, I think).  Is that correct?

And last, does Cisco support this notion of subnetting on its WAN
interfaces?  I haven't seen anything in the docs that indicates that they
don't, but experienced users would know for sure.

Email responses are preferred, but followups are ok.

Thanks

-- 

Mark T. Dornfeld, Cyantic Systems Corporation       Voice: (416) 621-6166
1 Eva Road Suite 301                            Facsimile: (416) 621-6212
Etobicoke, Ontario, M9C 4Z5 CANADA                  Email: mark@cyantic.com

-----------[000135][next][prev][last][first]----------------------------------------------------
Date:      Sun, 08 May 1994 14:51:30 GMT
From:      mark@cyantic.com (Mark T. Dornfeld)
To:        comp.protocols.tcp-ip
Subject:   Re: Subnet masks NOT on octet boundaries

In article <1994May08.135544.12171@cyantic.com> mark@cyantic.com (Mark T. Dornfeld) writes:
>I currently have the task of reconfiguration of three pairs of Cisco
>routers with each pair connecting a local and remote LAN.  TCP/IP and
>Ethertalk are the current protocols being routed.
>
>In order to conserve the Class C IP numbers that have been assigned, I
>would like to use just one Class C address for the WAN (the Cisco serial
>ports) and set up a subnet mask that uses part of the fourth octet. 
>I'll use one of the most common examples in the TCP/IP documentation and
>select two bits from the last octet to be part of the subnet mask. So the
>mask would look like this: 11111111.11111111.11111111.11xxxxxx if
>I understand the method correctly. So in decimal, the mask is
>255.255.255.192, correct?
>
>The assigned Class C network number (fictitious here) that I am working
>with is 192.139.228.  If the mask is correct above, then how are the subnet
>number and the node number expressed decimal for the purpose of
>configuration?
>
>I assume that I have the range of 1-191 for hosts and 192-223 for network
>numbers in the last octet (Numbers over 223 are not supposed to be used for
>network addresses, I think).  Is that correct?

This is a followup to my own folly.  If the mask is 255.255.255.192, then
there are four network numbers available: 0, 64, 128, 192.  I think this is
accurate.  Please comment if I am still off the wall.

By extension, with a mask of 255.255.255.224 there are eight network
numbers available: 0, 32, 64, 96, 128, 160, 192, and 224.

I am now guessing that the host numbers are additive such that for network
number 0, the host range is 1-31, for network number 32, the host range is
33-63 and so forth.  Is that now accurate?

>And last, does Cisco support this notion of subnetting on its WAN
>interfaces?  I haven't seen anything in the docs that indicates that they
>don't, but experienced users would know for sure.
>
>Email responses are preferred, but followups are ok.
>
>Thanks
>
>-- 
>
>Mark T. Dornfeld, Cyantic Systems Corporation       Voice: (416) 621-6166
>1 Eva Road Suite 301                            Facsimile: (416) 621-6212
>Etobicoke, Ontario, M9C 4Z5 CANADA                  Email: mark@cyantic.com


-- 

Mark T. Dornfeld, Cyantic Systems Corporation       Voice: (416) 621-6166
1 Eva Road Suite 301                            Facsimile: (416) 621-6212
Etobicoke, Ontario, M9C 4Z5 CANADA                  Email: mark@cyantic.com

-----------[000136][next][prev][last][first]----------------------------------------------------
Date:      8 May 1994 18:15:03 GMT
From:      tli@cisco.com (Tony Li)
To:        comp.protocols.tcp-ip
Subject:   Re: Subnet masks NOT on octet boundaries

In article <1994May08.145130.12730@cyantic.com> mark@cyantic.com (Mark T.
Dornfeld) writes: 
    >The assigned Class C network number (fictitious here) that I am working
    >with is 192.139.228.  If the mask is correct above, then how are the
 subnet 
    >number and the node number expressed decimal for the purpose of
    >configuration?
    >
    >I assume that I have the range of 1-191 for hosts and 192-223 for network
    >numbers in the last octet (Numbers over 223 are not supposed to be
 used for 
    >network addresses, I think).  Is that correct?
    
    This is a followup to my own folly.  If the mask is 255.255.255.192, then
    there are four network numbers available: 0, 64, 128, 192.  I think this is
    accurate.  Please comment if I am still off the wall.
    
Correct.  You don't want to use subnet number 0 or 192, if you can avoid it.

    By extension, with a mask of 255.255.255.224 there are eight network
    numbers available: 0, 32, 64, 96, 128, 160, 192, and 224.
    
    I am now guessing that the host numbers are additive such that for network
    number 0, the host range is 1-31, for network number 32, the host range is
    33-63 and so forth.  Is that now accurate?
    
Yup.

    >And last, does Cisco support this notion of subnetting on its WAN
    >interfaces?  I haven't seen anything in the docs that indicates that they
    >don't, but experienced users would know for sure.

Yes.

Tony

-----------[000137][next][prev][last][first]----------------------------------------------------
Date:      8 May 1994 19:18:08 GMT
From:      cricket@nsr.hp.com (Cricket Liu)
To:        comp.protocols.tcp-ip,comp.protocols.tcp-ip.domains
Subject:   Re: DNS and variable length subnets

Howard Hart (hch@hybrid.com) wrote:
: So if I understand this, it isn't possible to have two or more subdomains
: serving one Class C, subnetted or not?

No, what I said was that, in order to have multiple domains per class C,
you need to have one domain per IP address.  In other words, for domain
3.2.1.in-addr.arpa, you'd delegate to 4.3.2.1.in-addr.arpa,
5.3.2.1.in-addr.arpa, etc.

: I have what I thought was a 
: working model of this, two primaries, bcw.hybrid.com and dca.hybrid.com
: on one class C, with dca.hybrid.com being primary for the IN-ADDR
: record they shared. I am seeing problems though in that the 
: bcw.hybrid.com nameserver refuses to be authoritative, even for
: itself (goes all the way to BARRNET for name resolution). 
 
: Am I doing something that will never really work? Thanks in advance.

I'm afraid I don't have enough information on your setup to say.  Sounds
like you're saying that bcw.hybrid.com and dca.hybrid.com are both primary
master for hybrid.com, but that dca.hybrid.com is primary for the
in-addr.arpa domain that corresponds to Hybrid's network number.  If that's
the case, then yes, bcw may very well have to go all the way to the roots
occasionally for information in the in-addr.arpa domain.

--
cricket

cricket@hp.com / Hewlett-Packard Professional Services / Englewood, CO

-----------[000138][next][prev][last][first]----------------------------------------------------
Date:      8 May 1994 19:20:32 GMT
From:      cricket@nsr.hp.com (Cricket Liu)
To:        comp.protocols.tcp-ip
Subject:   Re: DNS and OpenView

Richard E. Blair (rblair@cabell.vcu.edu) wrote:

: Is anyone out there running HP OpenView on a Name Server?  We are
: running OpenView 3.2 on an HP 715/50 running HP-UX 9.01.  Whenever
: I start named and then run OpenView, I get the error message "ovw:
: unable to resolve "hostname" to get a license."  The name server
: itself work fine, and I believe it's configured properly.  What I
: don't know is exactly how OpenView attempts to resolve a hostname.

I've certainly done this before, and it may be that the licensing is
failing because you licensed OpenView back when you didn't have DNS
configured.  To check and see what OpenView's trying to do, turn on name
server debugging with "sig_named debug 1" and start OpenView.  Then check
/usr/tmp/named.run to see what it tried to look up.

--
cricket

cricket@hp.com / Hewlett-Packard Professional Services / Englewood, CO

-----------[000139][next][prev][last][first]----------------------------------------------------
Date:      9 May 1994 10:26:59 -0700
From:      Gary Palmer <pesky@scoob.xap.com>
To:        comp.sys.next.sysadmin,comp.sys.next.misc,comp.sys.mac.comm,comp.sys.mac.system,comp.protocols.tcp-ip,ca.unix
Subject:   Re: Setting up a home LAN with NEXSTEP/Unix

In article <2qipqu$m29@portal.gmu.edu>,
Christian Smith <csmith@blackplague.gmu.edu> wrote:
>Ok, here is what you should do.  Since the NeXT is using Slip to get to the
>outside world, it has TWO network interfaces, call them slip0 and en0.  Each
>of these should have its own ip address.  The slip0 interface will get its
>ip from the slip server you connect to.  The en0 interface will get its ip
>address from /etc/hostconfig.  Now, you have two choices

Erm, I dunno much about NeXT, but FreeBSD (and other Unixes) can
run quite happily with the SLIP i/f address set to the ethernet card
address, since routing is done by the SLIP's destination address,
not by it's local address (AFAIK). I can say this (with a bit of
uncertainty) as the machine I am sending this from has a SLIP i/f to
Netcom (designation 'sl0'), and a Ethernet i/f toa 10bT LAN (designated
'ed1'). Both have been ifconfiged to have the same local IP address, and a
default route has been set to the gateway machine at the other end of the SLIP
link, with the ifconfig for the ed1 card setting the route
to the LAn correctly. Unless NeXT does things differently, only
one IP address should be needed by the NeXT. Certainly,
2 (or more) IP addresses are needed for machines with multiple IP i/f's
of a more traditional nature (ethernet, FDDI, etc), but I think most SLIP
implimentations would fall over if the 'en0' and the 'sl0' were given
separate IP addresses.

*However* I could be wrong, but this setup works fine for us.

>Choice one is easy, choice two involves setting routed on the next, and getting
>you slip prrovider to set you up with a subdomaine, which they might or might
>not be willing to do.  I can't help you out to much with this one.

Again, AFAIK, routed isn't needed unless there are multiple possible
routes to the smae destination. All that is needed for SLIP is a default
route entry, made just after the SLIP interface is brought up.

Hope this help, and doesn't confuse anyone!

BTW: Can anyone give more details of the Nutshell routing book,
     e.g. ISBN, etc? Thanks

Gary

-----------[000140][next][prev][last][first]----------------------------------------------------
Date:      Mon, 9 May 1994 01:15:13 GMT
From:      smn@netcom.com (Subodh)
To:        comp.protocols.tcp-ip
Subject:   NOLINGER option for socket.

I have a small application in which the server that uses socket() takes whole 
lot of time when starting the server immediately after shuttin it down. 
Is there some option that I need to set when I create socket to avoid
this thing. I see this mainly on SGI machine. 

All help appriciated.
-Subodh

-----------[000141][next][prev][last][first]----------------------------------------------------
Date:      9 May 1994 12:01:27 -0600
From:      pashdown@xmission.com (Pete Ashdown)
To:        comp.protocols.tcp-ip
Subject:   Gateway trouble, puhleez help!

First off, is there a decent book or document that explains gateways and/or the
use of traceroute to resolve problems?

Second, here's my problem.  It seems that the companies who I purchased this
hardware from are slow to answer it, although it seems like a simple
configuration problem.

Here is the topology of my domain:

			Internet
			   |
			   |
			NAT LANB/290 (198.60.22.1)
			   |
			   |
Sparc (198.60.22.2)--------+-----Annex (198.60.22.6)
				   |   |   |   |   |
				   |   |   |   |   +-(198.60.22.100-115)
				   |   |   |   +-(192.146.183.0)
				   |   |   +-(199.165.191.0)
				   |   +-(199.165.252.0)
				   +-(199.165.251.0)

The problems are as such:

	1. The PPP/SLIP hosts outside my class C can reach the 198.60.22.0
           net with no problem.  Anywhere else does not work.
	2. The PPP/SLIP hosts inside the class C (100-115) can reach anywhere,
           but suffer from slow FTP transfers TO the Internet (~300 cps).
           However, sending FTP TO 198.60.22.2 is normal (~1400 cps).

Here are my gateways from config.annex (198.60.22.6, the SLIP/PPP server):

%gateway
annex 198.60.22.6

   # xmission.com
   net 198.60.22.0 gateway 198.60.22.6 metric 1 hardwired
   # iss.com
   net 192.146.183.0 gateway 192.146.183.80 metric 2 hardwired
   # satel.com
   net 199.165.191.0 gateway 199.165.191.1 metric 2 hardwired
   # stebbins.com
   net 199.165.252.0 gateway 199.165.252.5 metric 2 hardwired
   # slip.com
   net 199.165.251.0 gateway 199.165.251.51 metric 2 hardwired
   # Internet
   net default gateway 198.60.22.6 metric 2 hardwired

end


And 198.60.22.2's (the Sun server) gateways file:

# SLIP and PPP connections
host 198.60.22.100 gateway 198.60.22.6 metric 2 passive
host 198.60.22.101 gateway 198.60.22.6 metric 2 passive
host 198.60.22.102 gateway 198.60.22.6 metric 2 passive
host 198.60.22.103 gateway 198.60.22.6 metric 2 passive
host 198.60.22.104 gateway 198.60.22.6 metric 2 passive
host 198.60.22.105 gateway 198.60.22.6 metric 2 passive
host 198.60.22.106 gateway 198.60.22.6 metric 2 passive
host 198.60.22.107 gateway 198.60.22.6 metric 2 passive
host 198.60.22.108 gateway 198.60.22.6 metric 2 passive
host 198.60.22.109 gateway 198.60.22.6 metric 2 passive
host 198.60.22.110 gateway 198.60.22.6 metric 2 passive
host 198.60.22.111 gateway 198.60.22.6 metric 2 passive
host 198.60.22.112 gateway 198.60.22.6 metric 2 passive
host 198.60.22.113 gateway 198.60.22.6 metric 2 passive
host 198.60.22.114 gateway 198.60.22.6 metric 2 passive
host 198.60.22.115 gateway 198.60.22.6 metric 2 passive
host 198.60.22.200 gateway 198.60.22.6 metric 2 passive
host 198.60.22.201 gateway 198.60.22.6 metric 2 passive
host 198.60.22.202 gateway 198.60.22.6 metric 2 passive
#
# iss.com
net 192.146.183.0 gateway 198.60.22.6 metric 2 passive
# satel.com
net 199.165.191.0 gateway 198.60.22.6 metric 2 passive
# stebbins.com
net 199.165.252.0 gateway 198.60.22.6 metric 2 passive
# slip.com
net 199.165.251.0 gateway 198.60.22.6 metric 2 passive


The NAT LANB/290 has a setting for "static" gateways which only work for
networks.  I set them the same as the last four entries above.

Also, if I set the default gateway on the Annex to 198.60.22.1, the NAT can
not ping connected hosts outside the 198.60.22.0 class C.  If I set it as
above (198.60.22.6) it can ping them, but the aforementioned problems remain.

I messed around with traceroute today and it was inconclusive.  Any
suggestions at all on this would be great.  I would love to RTFM if I could
find the FM.  The FMs from NAT and Xylogics are terrible in this aspect.

-- 
			     My wall's vibrating

Pete Ashdown             pashdown@xmission.com            Salt Lake City, Utah
XMission - Internet Access For Utah - Data: 801 539 0900 - Voice: 801 539 0852

-----------[000142][next][prev][last][first]----------------------------------------------------
Date:      Mon, 9 May 1994 03:28:06 GMT
From:      sysseh@devetir.qld.gov.au (Steve Hocking)
To:        comp.protocols.tcp-ip
Subject:   Tcpdump & DLPI


	Has anyone ported tcpdump to use the DLPI interface? I'd like to
know before I think of tackling this one myself.

	Stephen


-----------[000143][next][prev][last][first]----------------------------------------------------
Date:      9 May 1994 11:48:50 -0400
From:      amol@ms.uky.edu (Amol Deshpande)
To:        comp.protocols.tcp-ip
Subject:   Re: Telnet Echo Option

donp@novell.com (don provan) writes:

>In article <roberto.2.00108F4B@jsb.co.uk> roberto@jsb.co.uk (Robert Osborne) writes:
>>Can anybody help me with a problem I am having with the Telnet echo option.
[ Description of echo problem deleted ]

>The manually disabled echo is probably being overridden by the OS when
>it disables the echo to protect your password.  Once the OS has the
>password, my guess is it "reenables" echoing, forgetting that echoing
>was initially disabled.

As I have mentioned before, I have the same problem. 
I tried to see if 'ed; would not echo (as suggested by someone). But the
result was the same.

As for the suggestion above, would not another IAC DONT ECHO turn echo off
again ? I tried to send an IAC DONT ECHO after login, but the server does not
acknowledge it and persists in echoing.

Is there a sequence in which the initial options should be negotiated ? Or
does any sequence work ?

thanks,

-amol
-- 
 Amol Deshpande , amol@ms.uky.edu
Curiosity killed the cat, but for a while I was a suspect.

-----------[000144][next][prev][last][first]----------------------------------------------------
Date:      9 May 1994 05:14:57 GMT
From:      blacey@cerf.net (Bruce B. Lacey)
To:        comp.client-server,comp.unix.osf.misc,comp.protocols.tcp-ip,comp.unix.programmer,comp.os.ms-windows.networking.tcp-ip,comp.os.ms-windows.programmer.networks
Subject:   Portable RPC package?

We are currently implementing a client/server signal processing 
workbench.  In pursuit of this formidable objective, we are in need of 
a portable remote-procedure call (RPC) package to handle the 
inter-process communication (IPC).  The client side of our application 
will run on Suns (Sun O/S 4.1.3, Solaris 1.0), Macs (System 7.0+), and 
PCs (Windows 3.1 or OS/2 2.1).  The server side will be restricted to 
Suns operating Sun O/S 4.1.3 or Solaris 1.0.  Does anyone know of a 
multi-platform package that meets these requirements?

Please respond via e-mail to blacey@cerf.net

Thanks in advance for your help.

+====================================================================+
| Bruce B. Lacey     Technical Group Manager of Systems Engnineering |
| XonTech, Inc.      blacey@cerf.net                                 |
| Van Nuys, CA 91406 (818) 787-7380 ext. 295                         |
 +~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~+ 
|     Specializing in massively parallel computing architectures.    |
+====================================================================+

-----------[000145][next][prev][last][first]----------------------------------------------------
Date:      9 May 94 13:43:48 -0500
From:      coreyt@vax1.mankato.msus.edu (Corey Thompson)
To:        comp.protocols.tcp-ip
Subject:   Re: Bargains on Comer's 3 Volume TCP/IP set

In article <1994May4.235045.14166@tron.bwi.wec.com>, carson@tron.bwi.wec.com (Dana Carson) writes:
> I got the Comer books before the first monthly mailer arrived.  Maybe
> they are having more of a response then they expected and there on back
> order?  You might want to check if they think that should have them
> already.
  
  Would someone repost the original message, or tell me where I can get
these 3 books?  Thank you.

-- 
Internet: coreyt@mankato.msus.edu |  Mankato State University
          coreyt%msus1@bitnet     |  Rocky Creek Software Development
Compu$seve: 71044,1260            |  Team OS/2

-----------[000146][next][prev][last][first]----------------------------------------------------
Date:      9 May 1994 08:08:25 GMT
From:      melbye@stkd20.alcatel.no (Jan O L Melbye)
To:        comp.protocols.tcp-ip
Subject:   Fastest way to re-connect?


Hi,

I have a client-server application running. The client produces a lot of data
that need to be stored when the communication for some reason goes down. To
avoid this problem I am trying to make the re-connection as fast as possible.
To be more specific: When the connection is resat and I try to connect to
the server again I get the "Address busy" error message.

Is there a really fast way to release the sockets in both ends? Could someone
send me / point me to a good example of a way to do this?

Please respond by e-mail to melbye@alcatel.no. (I will summarise back to this
group if I get any good answers)

-- Jan Olav

No silly disclaimer !  

-----------[000147][next][prev][last][first]----------------------------------------------------
Date:      Mon, 9 May 1994 13:43:25
From:      loss@husky.bloomu.edu (Doug Loss)
To:        comp.protocols.tcp-ip
Subject:   Re: Getting network name from IP address

In article <2ql02a$3ni@sgi.iunet.it> eb@iunet.it (Enrico Badella) writes:

>I'm wrting an application that needs to show the official network name
>given any IP address; for example, givven the IP address 18.172.1.2
>I would like to display mit.edu. Before hacking some code I played
>around with nslookup setting the querytype to ANY and asking for reverse
>addresses, 2.1.172.18.in-addr.arpa. (with the ending dot). The results
>were quite inconsistent since on some networks I would get a correct
>answer on others the answer was correct only if I identified a real
>subnet, other nameservers would simple reject the query.
 
>I have studied quite carefull name server configuration, nslookup manual
>and read the O'Reilly book on Bind&DNS. Are my failures due becauseI'm 
>trying to get nslookup do something strange or have I misunderstood something

   I don't know how helpful this will be, but a number of domains have 
neglected to or chosen not to register their reverse domains, which requires a 
separate registration from the standard domain.  That being the case, I'm 
afraid you're unlikely to be able to resolve every IP address to a name.

 
Doug Loss                        To stay young requires unceasing
Data Network Coordinator         cultivation of the ability
Bloomsburg University            to unlearn old falsehoods.
loss@husky.bloomu.edu
Voice (717) 389-4797

-----------[000148][next][prev][last][first]----------------------------------------------------
Date:      9 May 1994 09:00:17 GMT
From:      rparratt@london.micrognosis.com (Richard Parratt)
To:        comp.unix.programmer,comp.protocols.tcp-ip
Subject:   Re: IP address/sitename from network login

In article <grimesCpDv0q.1BJ@netcom.com>, grimes@netcom.com (George Grimes) writes:
|> In article <2qd5js$etj@tdc.dircon.co.uk> pmiles@tdc.dircon.co.uk (Peter Miles) writes:
|> >Can anyone suggest how, using C on a UNIX system (System V with sockets) 
|> >I can get details of the originating IP address (and hostname) of a 
|> >network login, AFTER the user has logged in using telnet or rlogin.
|> >
|> >Basically, I need a C program  which when run, returns the IP address of 
|> >the user's originating system.
|> >
|> >I know you can get this information with TCP wrapper, but that only works
|> >BEFORE the user has logged in -- I need this information after they've 
|> >logged in.
|> >
|> >If anyone has any ideas (or example code), I'd be extremely grateful.
|> >Please email replies, and I'll post a summary.
|> >
|> >		-- Pete
|> I wanted to do something like this to automatically set my DISPLAY environment
|> variable when I'm running X.  I used the ttyname() call to get the current tty
|> that I'm connected to then read the utmp file looking for the matching entry.
|> I do automatically in my .login file.
|> 
|> See "man ttyname" and "man utmp" for more information.  Send me email if  you
|> have specific questions.
|> 

On most machines "who am i" does the job (e.g. it tells you the hostname
you are logged in from, making it trivial to get the IP address).

Richard Parratt

-----------[000149][next][prev][last][first]----------------------------------------------------
Date:      9 May 1994 09:34:02 GMT
From:      eb@iunet.it (Enrico Badella)
To:        comp.protocols.tcp-ip
Subject:   Getting network name from IP address


Hello,

I'm wrting an application that needs to show the official network name
given any IP address; for example, givven the IP address 18.172.1.2
I would like to display mit.edu. Before hacking some code I played
around with nslookup setting the querytype to ANY and asking for reverse
addresses, 2.1.172.18.in-addr.arpa. (with the ending dot). The results
were quite inconsistent since on some networks I would get a correct
answer on others the answer was correct only if I identified a real
subnet, other nameservers would simple reject the query.

I have studied quite carefull name server configuration, nslookup manual
and read the O'Reilly book on Bind&DNS. Are my failures due becauseI'm 
trying to get nslookup do something strange or have I misunderstood something

Thanks  in advance to all who will help me.

================================================================================
Enrico Badella					email  softstar@pol88a.polito.it
Soft*Star s.r.l.				       eb@vax.cnuce.cnr.it
Via Camburzano 9				phone  +39-11-746092
10143 Torino, Italy				fax    +39-11-746487

	People are strange
	When you're a stranger	(J. Morrison)
================================================================================

-----------[000150][next][prev][last][first]----------------------------------------------------
Date:      Mon, 9 May 1994 14:59:08
From:      clarkb@netstar.com (Clark Bremer)
To:        comp.protocols.tcp-ip
Subject:   Re: Netware/IP product

In article <1994May5.150014.8408@aqm.com> rturner@qms.com (Randy Turner) writes:
>From: rturner@qms.com (Randy Turner)
>Subject: Netware/IP product
>Date: Thu, 5 May 1994 15:00:14 GMT


>I was under the impression that  Netware/IP from Novell was just an NLM that
>allowed
>you to tunnel netware packets through IP to take advantage of existing WAN IP
>networks.
 
>Is this true? or does it offer more functionality than just a router?

Novell does ship with a tcp/ip nlm that does add a tcp/ip stack when loaded.  
It doesn't do much unless you buy there NFS package that runs on top of it.  
This package turn the Netware file server into an NFS server, of course, but 
also provides print services (LPR/LPD type stuff) and other TCP/IP goodies.  
As far as I know (which ain't too far!) about the only thing the bare TCP/IP 
lets you do is network configuration stuff with SNMP.  I'm not sure if you 
need to purchase the NFS package to do IP Tunneling.  CB.


===========================================================================
          _  _               Clark Bremer     clarkb@netstar.com
         /  /_)              Software Engineer, NetStar Inc.
         \_/__)              10250 Valley View Road  MPLS, MN 55344

-----------[000151][next][prev][last][first]----------------------------------------------------
Date:      09 May 1994 15:32:26 GMT
From:      nelson@crynwr.crynwr.com (Russell Nelson)
To:        comp.protocols.tcp-ip
Subject:   Re: REQUEST: Packet Driver for a DEC Etherworks card.

In article <CpE7yI.Esr@curia.ucc.ie> dave@odyssey.ucc.ie writes:

   Can anyone point me to a packet driver (pktint or something ?) for
   a DEC etherworks 3 (turbo) card.

Digital is working on one.  I'm going to be testing it.  I don't know
what their release schedule is.  I'll announce its availability widely.

--
-russ <nelson@crynwr.com>      ftp.msen.com:pub/vendor/crynwr/crynwr.wav
Crynwr Software   | Crynwr Software sells packet driver support | ask4 PGP key
11 Grant St.      | +1 315 268 1925 (9201 FAX)    | Quakers do it in the light
Potsdam, NY 13676 | LPF member - ask me about the harm software patents do.

-----------[000152][next][prev][last][first]----------------------------------------------------
Date:      9 May 1994 16:32:00 GMT
From:      scotta@kije.gsfc.nasa.gov (Scott Austin)
To:        comp.protocols.tcp-ip
Subject:   Re: Bargains on Comer's 3 Volume TCP/IP set

Jeremy Elson <jelson@ren.psy.jhu.edu> wrote:
  
>Does anyone have a phone number for LCIS?  I'd really like to find out where
>my books are.  I, too, am beginning to have reservations.

Yes, it's 1-800-257-8345

Scott Austin
scotta@cnt.com




-----------[000153][next][prev][last][first]----------------------------------------------------
Date:      9 May 94 17:00:31 GMT
From:      manmetha@gauss.rutgers.edu (Rajesh Malhotra)
To:        comp.protocols.tcp-ip
Subject:   How does a server detect a client crash using sockets?


TCP Gurus,

	I have a traditional Client-Server application where the server
accept()s a connection and then forks a child to handle the request.
Now after the fork(), if the client crashed before doing a formal disconnect,
the child gets left behind and cleanup does not take place.

	How would the server (either the parent or the child process)
detect the clients crash?

	Any and all responses on this matter appreciated.

Thanx,

Raj.
-- 
===============================================================================
  __  __     .
 /   __/    /                               Raj Malhotra.
/   ( /_   /                                voice : 
        __/                                 email : manmetha@gauss.rutgers.edu

===============================================================================

-----------[000154][next][prev][last][first]----------------------------------------------------
Date:      9 May 1994 17:00:51 GMT
From:      dredwine@dow.com (David Redwine )
To:        comp.protocols.tcp-ip
Subject:   Duplicate IP address detection


   Does anyone have a simple way of detecting duplicate IP address usage on a LAN
!before! the offending node has done damage (like jamming NFS)?  It seems that most
PC IP implimentations do not check to see if the address you type in is already
in use. 

Thanks,
David Redwine  


-----------[000155][next][prev][last][first]----------------------------------------------------
Date:      9 May 1994 17:56:16 GMT
From:      mjo@iao.ford.com (Mike O'Connor)
To:        comp.protocols.tcp-ip
Subject:   Re: Getting network name from IP address

In article <2ql02a$3ni@sgi.iunet.it> eb@iunet.it (Enrico Badella) writes:

:I'm wrting an application that needs to show the official network name
:given any IP address; for example, givven the IP address 18.172.1.2
:I would like to display mit.edu. Before hacking some code I played
:around with nslookup setting the querytype to ANY and asking for reverse
:addresses, 2.1.172.18.in-addr.arpa. (with the ending dot). The results
:were quite inconsistent since on some networks I would get a correct
:answer on others the answer was correct only if I identified a real
:subnet, other nameservers would simple reject the query.
:
:I have studied quite carefull name server configuration, nslookup manual
:and read the O'Reilly book on Bind&DNS. Are my failures due becauseI'm 
:trying to get nslookup do something strange or have I misunderstood something
:
:Thanks  in advance to all who will help me.

Why not check the first octet?  If x.y.z.t in the range:

	0-127	do a whois -h rs.internic.net x.0.0.0
	128-191	do a whois -h rs.internic.net x.y.0.0
	192-255	do a whois -h rs.internic.net x.y.z.0

In your case, above, 18.172.1.2 would output something like:

Massachusetts Institute of Technology (NET-MIT-TEMP)
   1 Amherst Street
   Cambridge, MA 02139-1986

   Netname: MIT
   Netnumber: 18.0.0.0

   Coordinator:
      Schiller, Jeffrey I.  (JIS)  JIS@MIT.EDU
      (617) 253-8400

   Domain System inverse mapping provided by:

   STRAWB.MIT.EDU               18.71.0.151
   W20NS.MIT.EDU                18.70.0.160
   BITSY.MIT.EDU                18.72.0.3

   Record last updated on 14-Jan-94.

Am I missing something here?
						...Mike
-- 
 Michael J. O'Connor           |  Internet:  mjo@jobone.srl.ford.com
 Ford Motor Company, OPEO      |  UUCP:      ...!fmsrl7!opeo!mjo
 20000 Rotunda, Bldg. 1-3001   |  Phone:     +1 (313) 248-1260
 Dearborn, MI  48121           |  Fax:       +1 (313) 323-6277

-----------[000156][next][prev][last][first]----------------------------------------------------
Date:      9 May 1994 17:56:48 GMT
From:      kevinbr@dublin.sun.com (Kevin Brown - SunService Dubai)
To:        comp.protocols.tcp-ip
Subject:   Re: SLIP implementation on SUN OS

Hi 

PC NFS Contains the SunOS implementation of SLIP, versions for both 4.1.x and
2.x.......works fine.....

I am not sure of the license issues of this product :-)

Kevin


-----------[000157][next][prev][last][first]----------------------------------------------------
Date:      9 May 1994 19:04:55 +0200
From:      chris@ticsa.com (Chris Pinkham)
To:        comp.protocols.tcp-ip,comp.protocols.tcp-ip.ibmpc
Subject:   IP accounting program?

I'm looking for a program that will watch an Ethernet network and tell me
how many bytes passed to/from particular networks to/from other networks
in a measured period (e.g. one month/week/day etc.).

Is anyone doing this successfully? Free or commercial software is OK.

Thanks,
Chris

-----------[000158][next][prev][last][first]----------------------------------------------------
Date:      Mon, 9 May 1994 22:53:35 GMT
From:      wjb3@po.cwru.edu (Wes Barton)
To:        comp.protocols.tcp-ip
Subject:   Re: Bargains on Comer's 3 Volume TCP/IP set

In article <2qloi0$9pe@paperboy.gsfc.nasa.gov> scotta@kije.gsfc.nasa.gov (Scott Austin) writes:
>From: scotta@kije.gsfc.nasa.gov (Scott Austin)
>Subject: Re: Bargains on Comer's 3 Volume TCP/IP set
>Date: 9 May 1994 16:32:00 GMT
 
>Jeremy Elson <jelson@ren.psy.jhu.edu> wrote:
>  
>>Does anyone have a phone number for LCIS?  I'd really like to find out where
>>my books are.  I, too, am beginning to have reservations.
 
>Yes, it's 1-800-257-8345
 
>Scott Austin
>scotta@cnt.com

I missed the first part of this thread.

Did you have trouble with them?
I've had those books sitting on my shelf for months.
They've been relatively good with all the paperwork.

Meanwhile, Columbia House is all over my back.
Despite the fact that they are billing me for stuff I've never heard of
and never sent me what I ordered.

Back to the business at hand, I think that that is the correct phone number.

Wes



Wesley J. Barton
Dept of Computer Engineering & Science
Case Western Reserve University
Cleveland, Ohio, USA

-----BEGIN PGP PUBLIC KEY BLOCK-----
Version: 2.3

mQBNAi3MCLMAAAECAJiGZib53bngvsM8f/21bcm72GRblyHErhN8sOpWgYe+M9iY
0uwMz0sbGE2JsxJLdSr93Z4YKY4dlraDMfS9hUUABRG0JVdlc2xleSBKLiBCYXJ0
b24gPHdqYjNAcG8uY3dydS5lZHU+IEM=
=Rl+L
-----END PGP PUBLIC KEY BLOCK-----

-----------[000159][next][prev][last][first]----------------------------------------------------
Date:      Mon, 9 May 1994 23:10:15 GMT
From:      wjb3@po.cwru.edu (Wes Barton)
To:        alt.winsock,comp.protocols.tcp-ip,comp.protocols.tcp-ip.ibmpc
Subject:   Using TCP-IP software without a network

Hello,
I'm going to be away from my net access for a few months this summer.
But I need to be able to do some Winsock developement.  What I want to do is 
enable a loopback address.  I'm using Peter Tattam's Winsock.
What I need to know is if this can be done and if so how?
Can I just put 127.0.0.1 in the Winsock setup?
What can I do for a dummy Packet Driver? What about the Crynwr library?
Can I fake BOOTP?  How about DNS?

Thanks,
           Wes
Wesley J. Barton
Dept of Computer Engineering & Science
Case Western Reserve University
Cleveland, Ohio, USA

-----BEGIN PGP PUBLIC KEY BLOCK-----
Version: 2.3

mQBNAi3MCLMAAAECAJiGZib53bngvsM8f/21bcm72GRblyHErhN8sOpWgYe+M9iY
0uwMz0sbGE2JsxJLdSr93Z4YKY4dlraDMfS9hUUABRG0JVdlc2xleSBKLiBCYXJ0
b24gPHdqYjNAcG8uY3dydS5lZHU+IEM=
=Rl+L
-----END PGP PUBLIC KEY BLOCK-----

-----------[000160][next][prev][last][first]----------------------------------------------------
Date:      Mon, 9 May 1994 23:19:48 GMT
From:      rmadhok@anl.liv.usw.com (Raghu Madhok)
To:        comp.protocols.tcp-ip
Subject:   Re: Bargains on Comer's 3 Volume TCP/IP set

In article <2qfph8INNb7i@jhunix.hcf.jhu.edu> jelson@ren.psy.jhu.edu (Jeremy Elson) writes:
>Thayne Forbes <thayne@xmission.com> wrote:
>>:> : The Library of Computer and Information Sciences (LCIS), is currently
>>:> : having a promo with Comer's 3 books.  You can save some bucks, since the
>>:> : Publisher's prices on these is $164!  This offer is open to any
>>:> : non-members (wish *I* could).  The scoop is this:
 
>>:> The only problem is that I joined in March (maybe Feb) and I'm still waiting
>>:> for two of the Volumes
>>
>I'm in exactly the same position - I sent my little form in, they billed
>me for the five bucks (plus 8 bucks shipping), and I still haven't seen any
>books.  I sent the check to them at least 3-4 months ago.  They did send
 ...
>Does anyone have a phone number for LCIS?  I'd really like to find out where
>my books are.  I, too, am beginning to have reservations.

I ordered my books way back in Feb too, and I've got to say that I've
had no problems.  I received all three books soon after joining the club and
I've also spoken with them over the phone a couple of times and they have been
quite helpful.  Try the number 1-800-257-8345 and good luck.

...Raghu

netcomsv!anl!rmadhok

-----------[000161][next][prev][last][first]----------------------------------------------------
Date:      9 May 1994 23:23:16 GMT
From:      crhodes@cs.uq.oz.au (Colin Rhodes)
To:        comp.client-server,comp.unix.osf.misc,comp.protocols.tcp-ip,comp.unix.programmer,comp.os.ms-windows.networking.tcp-ip,comp.os.ms-windows.programmer.networks
Subject:   Re: Portable RPC package?

As an extension to Bruce's request, I'm working for a Brisbane based company
that will implementing a generic client server architecture between OS/2
(Presentation manager), and AIX3.2.4.  Since we could end up sending anything 
information off scanned cards to full graphics images, this will need to be
_really_ generic (ie. not just an ad hoc approach)  Later ports could take us
anywhere.

Three cannonical questions arise!

1.  Does anyone have any experience with doing something like this?  If so, 
are there any good resources that you used?

2.  Obviously we can choose between sockets, RPC, or something similar.  Again,
has anyone made this choice for client server stuff.  Why?

3.  Portability is an issue. Comments?

Any and all replies will be appreciated.  If interest is high I'll post a 
summary.

Ta all

Colin.
--
/.\./.\./.\./.\./.\./.\./.\./.\./.\./.\./.\./.\./.\./.\./.\./.\./.\./.\./.\./.\
Colin Rhodes                            These are my opinions, and do not 
email:  crhodes@uq.cs.oz.au             reflect the views of the SVRC. 
/.\./.\./.\./.\./.\./.\./.\./.\./.\./.\./.\./.\./.\./.\./.\./.\./.\./.\./.\./.\

-----------[000162][next][prev][last][first]----------------------------------------------------
Date:      10 May 1994 00:17:38 GMT
From:      maf@dunedin.acs.ohio-state.edu (Mark Fullmer)
To:        comp.protocols.tcp-ip
Subject:   Re: Duplicate IP address detection

In article <2qlq83$uq@raven.dow.com> Private_User@dow.com writes:
>
>   Does anyone have a simple way of detecting duplicate IP address usage on a LAN
>!before! the offending node has done damage (like jamming NFS)?  It seems that most
>PC IP implimentations do not check to see if the address you type in is already
>in use. 

Not before, but shortly after.  If the people using your network know you
will be knocking on their door if they do screwey thing like pick random
host IP addresses, problems like this tend to go away.

If you have a Unix box on your subnetwork that will run tcpdump, you
can use a utility called arpmon.  It creates a database of IP address <->
MAC address by looking at ARP broadcasts and optionally ARP replies.
It can optionally be configured to send you E-mail or do something else
when multiple MAC addresses are seen for a single IP address.

The output looks like:


Arpmon report started 0:15:8 5/8/94 as -k

IP: 0.0.0.0         Not-Resolved
     Hardware Address:         00:60:8C:C0:01:D9 (3Com (1990 onwards))
     Number of Arp broadcasts: 20
     First Seen:               15:9:57 3/16/94
     Last Seen:                15:10:19 3/16/94   Asking For 128.146.1.7

     Hardware Address:         08:00:09:63:B5:EE (Hewlett-Packard)
     Number of Arp broadcasts: 3
     First Seen:               15:18:28 3/16/94
     Last Seen:                15:24:35 3/16/94   Asking For 128.146.5.201

     Hardware Address:         02:60:8C:41:94:6C (3Com IBM PC; Imagen; Valid;
     Number of Arp broadcasts: 30
     First Seen:               15:41:14 4/12/94
     Last Seen:                15:41:51 4/12/94   Asking For 128.146.1.7

     Hardware Address:         08:00:07:E6:81:25 (Apple)
     Number of Arp broadcasts: 181
     First Seen:               15:47:56 3/15/94
     Last Seen:                20:20:4 4/7/94     Asking For 128.146.5.1

     Hardware Address:         08:00:07:E6:81:25 (Apple)
     Number of Arp Replies:    1
     First Seen:               17:47:8 3/23/94
     Last Seen:                17:47:8 3/23/94    Replying To aa:0:4:0:df:19

IP: 128.146.6.134   dunedin
     Hardware Address:         08:00:20:0A:01:A4 (Sun)
     Number of Arp Replies:    20848
     First Seen:               13:37:59 3/15/94
     Last Seen:                11:30:31 5/4/94    Replying To 8:0:20:8:23:3b

     Hardware Address:         08:00:20:0A:01:A4 (Sun)
     Number of Arp broadcasts: 475
     First Seen:               14:3:40 3/15/94
     Last Seen:                15:55:49 5/2/94    Asking For 128.146.6.18


It's available by anonymous ftp at curiosity.cob.ohio-state.edu:/pub/arpmon/*

--
mark
maf+@osu.edu

-----------[000163][next][prev][last][first]----------------------------------------------------
Date:      10 May 1994 07:23:45 -0400
From:      exe01568@char2.vnet.net (exe01568)
To:        comp.protocols.tcp-ip
Subject:   Re: Netware/IP product

clarkb@netstar.com (Clark Bremer) writes:

>In article <1994May5.150014.8408@aqm.com> rturner@qms.com (Randy Turner) writes:
>>From: rturner@qms.com (Randy Turner)
>>Subject: Netware/IP product
>>Date: Thu, 5 May 1994 15:00:14 GMT


>>I was under the impression that  Netware/IP from Novell was just an NLM that
>>allowed
>>you to tunnel netware packets through IP to take advantage of existing WAN IP
>>networks.
 
>>Is this true? or does it offer more functionality than just a router?
 
>Novell does ship with a tcp/ip nlm that does add a tcp/ip stack when loaded.  
>It doesn't do much unless you buy there NFS package that runs on top of it.  
>This package turn the Netware file server into an NFS server, of course, but 
>also provides print services (LPR/LPD type stuff) and other TCP/IP goodies.  
>As far as I know (which ain't too far!) about the only thing the bare TCP/IP 
>lets you do is network configuration stuff with SNMP.  I'm not sure if you 
>need to purchase the NFS package to do IP Tunneling.  CB.

This is incorrect, the ability to do IP tunneling is included with NetWare,
NFS is not required.  Also, it's not an NLM, it is used more like a lan driver.
Check Appendix A of the Novell NetWare v3.1[12] TCP/IP Transport Supervisor's
Guide.

--
Todd Campbell
exe01568@char2.vent.net
campbemt@ca6.uscourts.gov

-----------[000164][next][prev][last][first]----------------------------------------------------
Date:      Tue, 10 May 1994 01:41:26 GMT
From:      wdawson@willard.atl.ga.us (Willard Dawson)
To:        comp.unix.sys5.r3,comp.protocols.tcp-ip
Subject:   ISC 3.2, SLIP, and syslog

I've got an ISC SVR3 3.2 box.  Some time ago, I made this box a SLIP
bastion host for a LAN.  Thus, it has an SMC ethernet card (to our LAN)
and a SLIP interface through a 14.4 modem to the Internet.  The
"hostname" of the system is associated with the SLIP interface rather
than with the SMC card.  Perhaps that was a mistake?

Adding that SLIP interface caused all logging through syslog to stop.
However, I was able to re-enable logging by establishing a route from
one interface to the other.

So, recently I found it useful to have another SLIP interface on the
same system.  And, as expected, this also broke syslog.  However, I have
had no luck in creating route entries that fix the logging.

Is there some sort of useful rule that can be applied to cause syslog to
work in such a scenario?  Is this discussed in any known place?  I've
been meaning to check out O'Reilly's "Managing TCP/IP..." in hopes that
something was to be found therein... is it?

I just want my logging to work...  well, that and SLIP...

[Followups redirected.]

-----------[000165][next][prev][last][first]----------------------------------------------------
Date:      Tue, 10 May 1994 02:38:01 GMT
From:      longyear@netcom.com (Al Longyear)
To:        comp.protocols.tcp-ip
Subject:   Re: Getting network name from IP address

eb@iunet.it (Enrico Badella) writes:


>Hello,
 
>I'm wrting an application that needs to show the official network name
>given any IP address; for example, givven the IP address 18.172.1.2
>I would like to display mit.edu. Before hacking some code I played
>around with nslookup setting the querytype to ANY and asking for reverse
>addresses, 2.1.172.18.in-addr.arpa. (with the ending dot). The results
>were quite inconsistent since on some networks I would get a correct
>answer on others the answer was correct only if I identified a real
>subnet, other nameservers would simple reject the query.

If you are writing an application, then use the BSD procedure
"gethostbyaddr()". [Check your local manuals for the proper calling sequence.]

If the reverse name is not registered then you will get an error. Either 
accept the dotted IP address as the "name" or reject the connection.
-- 
Al Longyear           longyear@netcom.com

-----------[000166][next][prev][last][first]----------------------------------------------------
Date:      10 May 1994 05:08:54 GMT
From:      hkim@gradient.cis.upenn.edu (Hyogon Kim)
To:        comp.protocols.tcp-ip
Subject:   TCP retransmission question

Hi TCP gurus, I began to learn TCP through books and trying to understand
its behavior on a conceptual level but now I am confused about the roles 
of two retransmission algorithms...

Could someone please tell me what would happen to congestion window 
size and ssthresh if fast retransmission and timeout retransmission 
interfere with each other? I tried to make up a plausible scenario
below where those two algorithms ``take turns'' and mess up each
other's rationale in changing conwin and ssthresh. My question may
look stupid but please be kind and point out what's wrong in my
understading. Thanks a lot!

[for simplicity I assumed fixed packet(segment) size and the numbers are 
in terms of such packets]

===================================================================
time          ack sequence       conwin		ssthresh	
           ------------------------------------------------
t0                i               100
t1                i               100                               
t2                i               100                              
t3                i               103              50             
t4                i               104
t5  (i+1) timeout occurs...         1(?)           52(?)
t6                i                 2(?)
t7  (i+2) timeout                   1		    1(?)
t8                i                 2(?)
t9               (i+1)              1(?)
t10              (i+1)              2(?)
=================================================================== 

Starting from t1, 3 consecutive acks ask for packet i, so at the end of t3
the ssthresh is reduced to half the conwin, and conwin is increased
by 3 [fast retransmission]. 
   At t4 another ack arrives and it leads to the increase of conwin by 1[fast
retransmission]. 
   Then at t5, timeout occurs for packet (i+1). So the ssthresh is set to 
half the current conwin and conwin is reduced to 1 [timeout retransmission]. 
   Then another ack for packet i, resulting in the increase of conwin [fast
retransmission???]. 
   At t7, another timeout for (i+2) occurs, but this time reducing ssthresh 
to 1, which is the half the current conwin [timeout retransmission] . 
   At t9, (i+1) is acked at last and conwin is set to 1, no matter which 
algorithm is in effect.


So what's wrong?

-----------[000167][next][prev][last][first]----------------------------------------------------
Date:      10 May 1994 11:39:38 GMT
From:      spidey@rtp.vnet.ibm.com
To:        comp.protocols.tcp-ip,comp.protocols.tcp-ip.ibmpc
Subject:   Re: PD-RPC for DOS or Windows

In <2qcn6a$4eb@damon.irf.uni-dortmund.de>, bittner@.irf.uni-dortmund.de () writes:
>Hi Folks !
>
>I need a rpc-Version for DOS and/or for Windows in order to build
>rpc-clients AND rpc-server. Sun`s PC-NFS doesn't offer the rpc-server-mode.
>Are there solutions out there ? Public Domain or commercial ones ?
>
>Christian A. Bittner
>
>
>---
>Christian A. Bittner                     -------------------------------------------  eMail: 					   ---- "It's not a bug - it's a feature" ----
>bittner@marius.irf.uni-dortmund.de         ---- A programmer's famous last words  ----
>                                           -------------------------------------------


Christian,

You may want to look at Noblenet they offer a windows package that
is independent of the vendors TCP/IP it is a RPC library WINRPC.DLL
that sits on top of winsock compliant stacks.

They can be reached at 508-460-8222 ask for Fred Callahan.

Jim Sliwa

Disclaimer:This is not a product endorsement.
These views I express are those of myself and not my employer.


--- When the snake and the mongoose meet the mongoose prevails
---    For it is the jaws with which we speak, not the venom with
---    which we say it that determines our strength!



-----------[000168][next][prev][last][first]----------------------------------------------------
Date:      Tue, 10 May 1994 13:04:26 +0000
From:      kevinco@skutter3.london.silverplatter.com (Kevin Comer)
To:        comp.protocols.tcp-ip
Subject:   SO_KEEPALIVE for tli?

Is there an equivalent to the socket option SO_KEEPALIVE available
in the tli interface (t_optmgmt)? The tli documentation is hopelessly
non-specific.

-----------[000169][next][prev][last][first]----------------------------------------------------
Date:      Tue, 10 May 1994 13:11:44 GMT
From:      kkchin@dcs.warwick.ac.uk (K K Chin)
To:        comp.protocols.tcp-ip
Subject:   FAQ on PPP and SLIP !! PLEASE HELP !

Hi, I'm new to this news group, I like to know the basic different between
PPP and SLIP and the pro and cons of the them.

	Please direct me to FAQ ( I don't know where to find them ) if my 
question is already answer there ! Thanks !

kkchin
-- 

-----------[000170][next][prev][last][first]----------------------------------------------------
Date:      Tue, 10 May 1994 14:05:28 GMT
From:      rad@tyrell.net (Payless Store Systems)
To:        comp.protocols.tcp-ip
Subject:   Re: TCP/IP and X.25 over VSAT Sattelite

Frank Lofaro (ftlofaro@unlv.edu) wrote:
: In article <1994May6.212023.9340@tyrell.net> rad@tyrell.net (Bob Daniel) writes:
: >How does the Internet deal with the packet-per-character issue with telnet?
: >Are there more efficient ways to do telnet?
: >Are there X.25 terminal emulation programs for UNIX?
 
: Use telnet in line mode. It uses very little bandwidth.

We tried that. Our application does not like line mode.  How does the
Internet manage the telnet issue?

-----------[000171][next][prev][last][first]----------------------------------------------------
Date:      Tue, 10 May 1994 14:13:34 GMT
From:      rad@tyrell.net (Payless Store Systems)
To:        comp.protocols.tcp-ip
Subject:   Re: TCP/IP and X.25 over VSAT Sattelite

Frank Lofaro (ftlofaro@unlv.edu) wrote:

: Use telnet in line mode. It uses very little bandwidth.

I've tried this but have a problem with F-Keys.  It echoes the escape
sequence and <Enter> has to be pressed before it is processed.  Is there
a way around this?

If fkeys can be use in line mode, this would be a good solution for us.


--

bob.daniel
rad@tyrell.net

-----------[000172][next][prev][last][first]----------------------------------------------------
Date:      Tue, 10 May 1994 14:27:38 GMT
From:      surprenc@JSP.UMontreal.CA (Surprenant Colin)
To:        comp.protocols.tcp-ip
Subject:   Dynamic IP addr & mail

Hi,

I would like to know what are the possibilities for the delivery of mail
on systems that uses dialup accesses to the network + dynamic IP address 
allocation. 

One obvious solution would be to have a mail server on which users
which use dialup IP log in to read/write mail. 

Is there any other way?

--

| Colin Surprenant  | galaad@pubnix.qc.ca       |
| 514-849-0948      | ------------------------- |
| Montreal, Quebec  | surprenc@jsp.umontreal.ca |

     Real user of a Free UN!X - NetBSD.

-----------[000173][next][prev][last][first]----------------------------------------------------
Date:      10 May 94 20:14:14 CDT
From:      anh@kuhub.cc.ukans.edu
To:        comp.protocols.tcp-ip
Subject:   Re: How does a server detect a client crash using sockets?

In article <May.9.13.00.31.1994.3497@gauss.rutgers.edu>, manmetha@gauss.rutgers.edu (Rajesh Malhotra) writes:
> TCP Gurus,
> 
> 	I have a traditional Client-Server application where the server
> accept()s a connection and then forks a child to handle the request.
> Now after the fork(), if the client crashed before doing a formal disconnect,
> the child gets left behind and cleanup does not take place.
> 
> 	How would the server (either the parent or the child process)
> detect the clients crash?
> 
> 	Any and all responses on this matter appreciated.
> 
> Thanx,
> 
> Raj.

When the client goes down, the server will be signaled there is something
on the socket, but upon reading the socket there will be nothing to read.
Eh, clear? :-)


select(magic thunderbolt,(struct timeval*)0); /* Wait indefinately for io*/
n=read(zeus socket);
if(n==0) {
	printf("Client is down\n");
}


That is, when select() returns, and there is no error, and there is no timeout and a
read() yields nothing the client has died.

Naturally it works the other way too. The client uses the same technique to find out if
the server died.


Anh

-----------[000174][next][prev][last][first]----------------------------------------------------
Date:      10 May 94 16:41:55 GMT
From:      manmetha@gauss.rutgers.edu (Rajesh Malhotra)
To:        comp.protocols.tcp-ip
Subject:   Re: How does a server detect a client crash using sockets?

In article <May.9.13.00.31.1994.3497@gauss.rutgers.edu>, manmetha@gauss.rutgers.edu (Rajesh Malhotra) writes:
> 
> TCP Gurus,
> 
> 	I have a traditional Client-Server application where the server
> accept()s a connection and then forks a child to handle the request.
> Now after the fork(), if the client crashed before doing a formal disconnect,
> the child gets left behind and cleanup does not take place.
> 
> 	How would the server (either the parent or the child process)
> detect the clients crash?
> 
> 	Any and all responses on this matter appreciated.
> 
> Thanx,
> 


	As a followup, how does so_keepalive work? I set this option
on the socket and the documentation says that a SIGPIPE should be returned
to the process associsted with the process, but I do not seem to get it after
the client has crashed. Does my server application have to send a periodic
transmission on the socket, or is that handled by the lower layers?

	Any and all responses appreciated. I am running this application
on an IBM RS6000 running AIX 3.2.4

Thanx,

Raj.
-- 
===============================================================================
  __  __     .
 /   __/    /                               Raj Malhotra.
/   ( /_   /                                voice : 
        __/                                 email : manmetha@gauss.rutgers.edu

===============================================================================

-----------[000175][next][prev][last][first]----------------------------------------------------
Date:      Tue, 10 May 1994 17:47:18 GMT
From:      donp@novell.com (don provan)
To:        comp.protocols.tcp-ip
Subject:   Re: Netware/IP product

In article <1994May5.150014.8408@aqm.com> rturner@qms.com writes:
>I was under the impression that Netware/IP from Novell was just an NLM
>that allowed you to tunnel netware packets through IP to take
>advantage of existing WAN IP networks.
>
>Is this true? or does it offer more functionality than just a router?

NetWare/IP is just what it sounds like: a product which allows you to
run the NetWare protocols over a TCP/IP network. For that reason, it has
features which allows the type of simple configuration and automated
service discovery that NetWare users are familiar with.

In order to support various migration paths, NetWare/IP also allows
you to connect new NetWare/IP networks together with old, established
NetWare/IPX networks.  You could use this capability to connect two or
more NetWare/IPX installations by using a NetWare/IP network. One
might call this tunnelling, but the NetWare/IP people put a lot of
ease-of-use features into their product, so they'd probably consider
it a lot more.

For simple tunnelling, you can use the IPTunnel NetWare OS LAN driver
which has been included in every release of NetWare since v3.11. This
requires manual configuration. You have to go to each node on the
tunnel and configure it to know about all the other nodes on the
tunnel, for example. But if your needs are simple enough and don't
change very often, this will get the job done.

						don provan
						donp@novell.com

-----------[000176][next][prev][last][first]----------------------------------------------------
Date:      Tue, 10 May 1994 19:07:31 GMT
From:      donp@novell.com (don provan)
To:        comp.protocols.tcp-ip
Subject:   Re: Telnet Echo Option

In article <2qlm12$dl6@s.ms.uky.edu> amol@ms.uky.edu (Amol Deshpande) writes:
>donp@novell.com (don provan) writes:
>>The manually disabled echo is probably being overridden by the OS when
>>it disables the echo to protect your password.  Once the OS has the
>>password, my guess is it "reenables" echoing, forgetting that echoing
>>was initially disabled.
>
>As for the suggestion above, would not another IAC DONT ECHO turn echo off
>again ? I tried to send an IAC DONT ECHO after login, but the server does not
>acknowledge it and persists in echoing.

Could be a problem with the coupling of the OS and the Telnet support.
If the OS turns its own echoing on without telling Telnet, the Telnet
support is will act as you describe. As far as it's concerned, the
state is still IAC WONT ECHO, so, as per the RFCs, it would ignore any
attempts to renegotiate to the existing state. Try negotiating IAC DO
ECHO and then IAC DONT ECHO and see if that convinces the Telnet
support to get the OS echoing back in line.

As I tried to say in my initial note, the basic problem you've
discovered is that this Telnet Server does not support clients doing
remote echoing. That could be considered either a bug or a missing
feature, but in either case, to do what you're trying to do, it sounds
like someone will have to do a little work on the software.  Even if
you find a way to correct the echoing after login, there will, no
doubt, be other applications later that mess around with the echoing.
Once they do, you'll be in the same boat.
						don provan
						donp@novell.com

-----------[000177][next][prev][last][first]----------------------------------------------------
Date:      Tue, 10 May 1994 19:47:54 GMT
From:      rstevens@noao.edu (W. Richard Stevens)
To:        comp.protocols.tcp-ip
Subject:   Re: How does a server detect a client crash using sockets?

> As a followup, how does so_keepalive work?

Check out my recent book "TCP/IP Illustrated" (Addison-Wesley, 1994).
There's a complete chapter with all the gory details of the keepalive
option.  Also check out Appendix E of this book for common configuration
parameters for the keepalive option, since everyone's first question
after learning what it does is "how can I set the keepalive time to
some value less than 2 hours."

> I set this option
> on the socket and the documentation says that a SIGPIPE should be returned
> to the process associsted with the process, but I do not seem to get it after
> the client has crashed. Does my server application have to send a periodic
> transmission on the socket, or is that handled by the lower layers?
> 	Any and all responses appreciated. I am running this application
> on an IBM RS6000 running AIX 3.2.4

Your documentation is basically wrong.  TCP generates SIGPIPE when an
RST is received on a connection--you're writing to a connection that's
been reset by the other end.  TCP does *not* generate this when your
end receives a FIN, due to TCP's half-close feature.

How to detect that the client has "crashed" requires that you be more
specific with what "crash" means.  If the client *process* crashes, its
TCP should send a FIN, which delivers an EOF to your end of the connection.
If the client *host* crashes, a keepalive probe can detect this, though
the error returned to your end of the connection depends whether the
client host has rebooted or not when the keepalive probes are sent.  If
the client *host* crashes and the connection is idle (you don't try to
send something to the client) and the keepalive option is not set, your
end will sit there, and sit there, and sit there ...

	Rich Stevens

-----------[000178][next][prev][last][first]----------------------------------------------------
Date:      Tue, 10 May 1994 20:08:06 GMT
From:      wjb3@po.cwru.edu (Wes Barton)
To:        comp.protocols.tcp-ip
Subject:   Re: Bargains on Comer's 3 Volume TCP/IP set

>In article <2qfph8INNb7i@jhunix.hcf.jhu.edu> jelson@ren.psy.jhu.edu (Jeremy 
 Elson) writes:>>Thayne Forbes <thayne@xmission.com> wrote:
>>>:> : The Library of Computer and Information Sciences (LCIS), is currently
>>>:> : having a promo with Comer's 3 books.  You can save some bucks, since the
>>>:> : Publisher's prices on these is $164!  This offer is open to any
>>>:> : non-members (wish *I* could).  The scoop is this:
 
>>>:> The only problem is that I joined in March (maybe Feb) and I'm still waiting
>>>:> for two of the Volumes
>>>
>>I'm in exactly the same position - I sent my little form in, they billed
>>me for the five bucks (plus 8 bucks shipping), and I still haven't seen any
>>books.  I sent the check to them at least 3-4 months ago.  They did send
 ...
>>Does anyone have a phone number for LCIS?  I'd really like to find out where
>>my books are.  I, too, am beginning to have reservations.
 
>I ordered my books way back in Feb too, and I've got to say that I've
>had no problems.  I received all three books soon after joining the club and
>I've also spoken with them over the phone a couple of times and they have been
>quite helpful.  Try the number 1-800-257-8345 and good luck.
 
>...Raghu
 
>netcomsv!anl!rmadhok
I got the bill and the books at the same time.  

Wes
Wesley J. Barton
Dept of Computer Engineering & Science
Case Western Reserve University
Cleveland, Ohio, USA
wjb3@po.cwru.edu

-----------[000179][next][prev][last][first]----------------------------------------------------
Date:      10 May 1994 21:38:24 GMT
From:      learned@winternet.com (Ed Learned)
To:        comp.protocols.tcp-ip
Subject:   Newbie Question

I've just set up a TCP environment including a few rs/6000's and os/2
workstations, running IBM's TCP/IP. I would like to set up a mail
service to allow me to get my mail on my OS/2 workstation, rather than 
on my account on one of the 6000's. What do I need to look into  
to accomplish this?

--

    Ed Learned                              |  Information
    Ed.Learned@mmbbs.mn.org                 |  Highway
    learned@winternet.com                   |  Worker

-----------[000180][next][prev][last][first]----------------------------------------------------
Date:      10 May 1994 23:36:51 GMT
From:      dimitrov@lab.ultra.nyu.edu (Isaac Dimitrovsky)
To:        comp.unix.sys5.r3,comp.protocols.tcp-ip
Subject:   Re: SOLVED: Strange behaviour of TCP/IP in ISC 4.0 - BUT...

In article <2qdbol$ip6@opel.secondsource.com> johnk@secondsource.COM (John Kennedy) writes:
>In article <2qb0uo$jgc@euas20.eua.ericsson.se> per@erix.ericsson.se (Per Hedeland) writes:
>>                      ISC->ISC     ISC->Sun     Sun->ISC     Sun->Sun
>> nbuf  buflen           secs         secs         secs         secs
>>====== ======           ====         ====         ====         ====
>> 65536     16            42           35            9           13 
>> 32768     32            23           18            5            7 
>> 16384     64            13           10            4            4 
>>  8192    128             7            6            3            2 
>>  4096    256             4            4            2            2 
>>  2048    512             9            3            3            1 
>>  1024   1024             4            2            2            0 
>>   512   2048            14 !          2            2            0 
>>   256   4096            51 !!         2            2            0 
>>   128   8192            50 !!         2            2            0 
>>    64  16384            51 !!         2            2            0 
>>    32  32768            50 !!         2            2            0 
>>    16  65536            50 !!         2            2            0 
>>
>
>Seems that when buflen > MTU, and the IP level has to generate multiple
>packets, things go downhill.  Check with netstat -i for the MTU for
>your particular device.  (Can that be changed?)  You may be able
>to decrease that value and see if the performance drops off at
>a different point.
>
>-- 
>John Kennedy                     johnk@secondsource.com 
>Second Source, Inc.
>Annapolis, MD

If this problem could be merely attributed to fragmentation from the ISC sender,
you would also see slowdown in the ISC->Sun figures.  Even if reassembly is much
worse on the ISC, you'd see something on the Sun. No?

Wayne Berke


-----------[000181][next][prev][last][first]----------------------------------------------------
Date:      11 May 94 08:41:01 PDT
From:      rkobenter@vmsmail.gov.bc.ca
To:        comp.protocols.tcp-ip,comp.protocols.tcp-ip.ibmpc
Subject:   Re: PD-RPC for DOS or Windows

In article <2qnrpq$lul@sernews.raleigh.ibm.com>, spidey@rtp.vnet.ibm.com writes:
> In <2qcn6a$4eb@damon.irf.uni-dortmund.de>, bittner@.irf.uni-dortmund.de () writes:
>>Hi Folks !
>>
>>I need a rpc-Version for DOS and/or for Windows in order to build
>>rpc-clients AND rpc-server. Sun`s PC-NFS doesn't offer the rpc-server-mode.
>>Are there solutions out there ? Public Domain or commercial ones ?
>>
>>Christian A. Bittner
>>
>>
>>---
>>Christian A. Bittner                     -------------------------------------------  eMail: 					   ---- "It's not a bug - it's a feature" ----
>>bittner@marius.irf.uni-dortmund.de         ---- A programmer's famous last words  ----
>>                                           -------------------------------------------
> 
> 
> Christian,
> 
> You may want to look at Noblenet they offer a windows package that
> is independent of the vendors TCP/IP it is a RPC library WINRPC.DLL
> that sits on top of winsock compliant stacks.
> 
> They can be reached at 508-460-8222 ask for Fred Callahan.
> 
> Jim Sliwa
> 
> Disclaimer:This is not a product endorsement.
> These views I express are those of myself and not my employer.
> 
> 
> --- When the snake and the mongoose meet the mongoose prevails
> ---    For it is the jaws with which we speak, not the venom with
> ---    which we say it that determines our strength!
> 
> 
The original request was for a PDversion, currently only SPRY seems to have put
their TI/RPC module into the PD  (ask info@spry.com)
as for the other mentioned ones here, there are a number of 'licenced'
versions of TI/RPC and ONC/RPC.  I saw a posting from Frontier pushing theirs
you can compare this to NetMnages' or FTP Software Inc. or others..

Smeone mentioned NobleNet.. I also looked at that and really liked it
by now they have a WinSock version available. The NetWise product is 
also interesting, I believe they support TI/RPC.

Hope this helps

--not affiliated with any of the vendors mentioned here--

robert kobenter
speaking only for myself and not the posting organization
rwkobent@bcsc02.gov.bc.ca




-----------[000182][next][prev][last][first]----------------------------------------------------
Date:      11 May 1994 18:40:13 -0700
From:      skl@Connectivity.com (Samuel Lam)
To:        comp.protocols.tcp-ip,comp.sys.novell,bit.listserv.novell
Subject:   Re: Subnetting and Netware

In article <1994May11.203948.27795@cherokee.nsuok.edu>,
 landin@cherokee.nsuok.edu (Mark Landin) wrote:
>My question is whether another server on the backbone (call it S3) will be
>able to figure out that to get to 111.111.111.2 you need to go through server
>S1, but that 111.111.111.254 is accessible only through server S2.
>
>I would try this, but the servers in question are inconveniently far apart,
>and this looks like a pretty simple "yes-or-no" proposition. 

The simple answer is "no".  To make it work you would need to either
1) use two unique class C IP network numbers on the two stub networks
or 2) use a third subnet of the non-contiguous class C network (the one
used for the stubs) on the backbone network.

...Sam
-- 
<skl@Connectivity.com> -- Connectivity Technology Inc.

"Packet Driver, WinSock & TCP/IP CD-ROM" product information:
{gopher,www,ftp}.CDPublishing.com or <info@CDPublishing.com>


-----------[000183][next][prev][last][first]----------------------------------------------------
Date:      Wed, 11 May 1994 17:28:07 -0500
From:      johns@oxygen.house.gov (John Schnizlein)
To:        comp.protocols.tcp-ip
Subject:   Re: Subnet masks NOT on octet boundaries

In article <2qja77$fus@cronkite.cisco.com>, tli@cisco.com (Tony Li) wrote:

> In article <1994May08.145130.12730@cyantic.com> mark@cyantic.com (Mark T.
> Dornfeld) writes: 
 [lots of correct stuff]     
>     I am now guessing that the host numbers are additive such that for network
>     number 0, the host range is 1-31, for network number 32, the host range is
>     33-63 and so forth.  Is that now accurate?
>     
> Yup.
> Tony

Hold on to that Yup, please.
Remember that using subnet 0 is problematic, so hosts of ...1 - ...31 are
out.
For the first usable subnet x.x.x.32, the last value of 63 has all ones in
the
host part, so it could be mistaken for a directed broadcast.
The best rule to avoid problems is avoid zero and all-ones in both subnet
and
host fields.

Tony, did I learn it wrong, or are the rules changed?
-- 
Badges! we don't need no stinking badges!         |
disclaimer! we don't need no stinking disclaimer! | John M. Schnizlein
everybody knows nobody can represent the views of | johns@oxygen.house.gov
435 elected policy makers.                        | router jockey

-----------[000184][next][prev][last][first]----------------------------------------------------
Date:      Wed, 11 May 1994 11:09:18 GMT
From:      johnston@hookup.net (Stewart Johnston)
To:        comp.protocols.tcp-ip
Subject:   TCP/IP   D.E. Comer Volume III

TCP/IP   D.E. Comer Volume III

I have just purchased "Internetworking with TCP/IP volume III" by Douglas E. Comer and David
L. Stevens, published by Pentice Hall.  On the back of the book there is a note that reads:
	"  - Includes programs available from the publisher and by FTP. "

I assume this means the example code in this book is available through FTP.  Unfortunately the
authors failed to publish the FTP sight address.

Could someone please help me find the FTP address for the above source code.

Your help would be greatly appreciated.

Thanks
Stewart Johnston
johnston@hookup.net

-----------[000185][next][prev][last][first]----------------------------------------------------
Date:      11 May 1994 11:30:48 GMT
From:      al@runit.sintef.no (Arne Langmo)
To:        comp.protocols.tcp-ip,comp.dcom.sys.wellfleet,comp.dcom.sys.cisco
Subject:   Re: Variable IP subnets with RIP on Wellfleet

In article <2q1f9m$kjc@paperboy.wellfleet.com>,jliv@wellfleet.com writes:
|> In article <2pjpq6$3tk@agate.berkeley.edu>
|> sklower@oboe.CS.Berkeley.EDU (Keith Sklower) writes:
|> 
|> > It would be good if somebody from wellfleet or
|> > cisco could authoritatively say whether their products support RIP version > 2.
|> 
|> We do not support RIPv2, at this time.

But has Wellfleet any support for variable IP subnet mask (CIDR) with OSPF?
If you do, from what version?

We're on 5.77e, and get this mess.:

I 05/11/94 13:16:53 ospf: 'Invalid IP mask, 193.219.32.0/255.255.224.0 ignored'
I 05/11/94 13:16:53 ospf: 'Invalid IP mask, 193.69.0.0/255.255.0.0 ignored'
I 05/11/94 13:16:53 ospf: 'Invalid IP mask, 193.71.0.0/255.255.0.0 ignored'
I 05/11/94 13:16:53 ospf: 'Invalid IP mask, 193.90.0.0/255.255.0.0 ignored'


Arne Langmo                          E-Mail: Arne.Langmo@runit.sintef.no
SINTEF Runit                                        Phone: +47 7359 2068
N-7034 Trondheim                                      FAX: +47 7359 1700
NORGE                                                             NORWAY

-----------[000186][next][prev][last][first]----------------------------------------------------
Date:      Wed, 11 May 1994 11:50:36 GMT
From:      zhagg@zh014.ubs.ubs.ch (Robert Agg)
To:        comp.protocols.tcp-ip
Subject:   WANTED: TCP/IP NFS for WINDOWS from Frontier Technologies Corp.

Hi world,

I am looking for a) the address of Frontier Technologies Corp. 
                 b) the name and address of a dealer close to or in Switzerland
                   
to obtain their product TCP/IP NFS for Windows.

Any pointers please via E-Mail.

Thanks
Robert 




-----------[000187][next][prev][last][first]----------------------------------------------------
Date:      Wed, 11 May 1994 11:53:11 GMT
From:      coert@athena.research.ptt.nl (Vonk C.J.S.)
To:        comp.protocols.tcp-ip
Subject:   BSD syslog(d), is there a RFC for this, is there an M$Windows port

Is the BSD syslogd protocol specified?  I went through the rfc-index.txt,
but couldn't find it.  And is there a WinSock port of this deamon that
I might run under Windows NT.  (yes, I know I might use the event deamon,
but I'm developing for Open VMS).

				coert


.............................................................................
Coert J. Vonk                                          voice: +31 70 332.3614
                                                         fax: +31 70 332.7807
C.J.S.Vonk@research.ptt.nl                             where: the Netherlands
X.400:/C=NL/ADMD=400NET/PRMD=PTT Research/O=PTT Research/S=Vonk/G=Coert/I=CJS
.............................................................................
programming is like sex - one mistake and you support it a lifetime   - sears
-- 
Coert J. Vonk                                          voice: +31 70 332.3614
                                                         fax: +31 70 332.7807
C.J.S.Vonk@research.ptt.nl                             where: the Netherlands

-----------[000188][next][prev][last][first]----------------------------------------------------
Date:      Wed, 11 May 1994 13:21:56 GMT
From:      alfa@cix.compulink.co.uk ("Alfa Systems")
To:        comp.protocols.tcp-ip
Subject:   Re: Portable RPC package?

Colin Rhodes wrote:

> Any and all replies will be appreciated.  If interest is high I'll post
> a summary.

Please do. I would be most interested to know what you find out.

Kin Ming Looi
Alfa Systems Limited
alfa@cix.compulink.co.uk

-----------[000189][next][prev][last][first]----------------------------------------------------
Date:      Wed, 11 May 1994 13:32:29 GMT
From:      barnett@grymoire.crd.ge.com (Bruce Barnett)
To:        comp.protocols.tcp-ip
Subject:   Overhead measurement

Last night I read an article I wanted to reply to, but this morning
the article expired. Sorry I cannot provide a reference.

Someone asked about a way of finding out the overhead and efficiency
of an FTP connection. There are a lot of variables:
	Buffer sizes (24KB is used on SunOS).
	TCP Implementation details
	Speed of processor
	Whether the Binary or ASCII FTP option is set.
	Disk overhead

The later is quite significant. Using "Binary" mode can double the
transfer rate. 

There is no simple rule that can be applied to all implementations,
but I presented a paper at InterOp last week describing a technique I
use to break down elapsed time of a TCP into categories. For instance,
I used "ttcp", 32K Buffers, to send a 67MByte file between two
SPARC-10's running SunOS 4.1.3. It was a memory-to-memory transfer (no
disk). The elapsed time was 62.54 seconds, and the transfer rate was
1.087 mBytes/second. The breakdown was:

What				% of Time	% of Packets	Ave. Gap
---
Transmitting			88.93		100		*
---
Packet gap within 32K buffer	2.71		90.16		40 usec.
Packet gap after 32K buffer	0.45		2.15		280 usec.
Waiting for Receive Window to empty:
				7.01		3.46		2730 usec.
etc.
---

There were 16 different "states", of which the above is the most
significant. The paper discusses in detail 4 variations. Using 4K
buffers increases the time waiting for an empty buffer to 33% of
total time. A Disk-to-Memory transfer increases the inter-packet delay
to 7.8%, and the delay after the buffer increases to 6.5% of the time.

Using a SPARC-2 with SunOS 4.1.1 as a source shows that 51% of the
time is spent doing retransmissions.

As I said, it depends. If you want a copy of the paper by e-mail, let
me know...
--
Bruce Barnett <barnett@crd.ge.com> uunet!crdras!barnett

-----------[000190][next][prev][last][first]----------------------------------------------------
Date:      11 May 1994 13:48:20 GMT
From:      spidey@rtp.vnet.ibm.com
To:        comp.client-server,comp.unix.osf.misc,comp.protocols.tcp-ip,comp.unix.programmer,comp.os.ms-windows.networking.tcp-ip,comp.os.ms-windows.programmer.networks
Subject:   Re: Portable RPC package?

In <2qmgl4$phg@uqcspe.cs.uq.oz.au>, crhodes@cs.uq.oz.au (Colin Rhodes) writes:
>As an extension to Bruce's request, I'm working for a Brisbane based company
>that will implementing a generic client server architecture between OS/2
>(Presentation manager), and AIX3.2.4.  Since we could end up sending anything 
>information off scanned cards to full graphics images, this will need to be
>_really_ generic (ie. not just an ad hoc approach)  Later ports could take us
>anywhere.
>
>Three cannonical questions arise!
>
>1.  Does anyone have any experience with doing something like this?  If so, 
>are there any good resources that you used?
>
>2.  Obviously we can choose between sockets, RPC, or something similar.  Again,
>has anyone made this choice for client server stuff.  Why?
>
>3.  Portability is an issue. Comments?
>
>Any and all replies will be appreciated.  If interest is high I'll post a 
>summary.
>
>Ta all
>
>Colin.

Colin,   Opinions to your questions:

1) Any experience.   Not sure if this helps I am not doing graphics specifically
   but have done Multi client with Multi server implementation with data ranges
   around 8k of mixed format data.   With respect to RPC programming you may
   want to look at "Power programming in RPC" as a good starter.   They may
   even have an example to help you out.

2) We chose SUN RPC versus sockets and here are the reasons
    - RPC provides routines for doing the necessary byte swapping, character
      conversion and portability we needed since the clients and servers
      were not all intel machines, not all ascii and some were windows.

    - RPC provided an easy interface for setting up and managing the connections
      as well as locating the server for the function in a network.

3) We initially had a portability issue with the multitude of TCP stacks
   available for Windows.  The code we had written was portable from to
   Unix, OS/2 and other environments but on Windows we had differences for
   the vendor stacks.   We ended up using a WINRPC.DLL available from Noblenet
   that provided access to all the RPC functions we used and sits on top
   of WINSOCK.   This made the code portable with respect to the RPC aspects
   by isoalting against system calls.

Jim Sliwa

Disclaimer:   The opinions expressed are mine and not those of my employer.
Any reference to products is not an endorsement by myself or my employer.


--- When the snake and the mongoose meet the mongoose prevails
---     For it is the jaws with which we speak 
---     not the venom with which we say it
---     that determines our strength!




-----------[000191][next][prev][last][first]----------------------------------------------------
Date:      11 May 94 19:05:04
From:      drw@kutta.mit.edu (Dale R. Worley)
To:        comp.protocols.tcp-ip
Subject:   Re: Duplicate IP address detection

In article <2qmjr2$49b@charm.magnus.acs.ohio-state.edu> maf@dunedin.acs.ohio-state.edu (Mark Fullmer) writes:
   In article <2qlq83$uq@raven.dow.com> Private_User@dow.com writes:
   >
   >Does anyone have a simple way of detecting duplicate IP address usage on a LAN
   >!before! the offending node has done damage (like jamming NFS)?  It seems that most
   >PC IP implimentations do not check to see if the address you type in is already
   >in use. 

   Not before, but shortly after.  If the people using your network know you
   will be knocking on their door if they do screwey thing like pick random
   host IP addresses, problems like this tend to go away.

Especially if you tend to commit physical violence upon the offenders.

But basically, there's no way (within IP) to tell if a particular
address is in use, except by attempting to send to it.  However, that
requires that the sender have an IP address, to which the reply can be
returned.

Dale

Dale Worley		Dept. of Math., MIT		drw@math.mit.edu
--
"If you could have any amount of money... How much would you want?"
"All of it."
-- "Cerebus"

-----------[000192][next][prev][last][first]----------------------------------------------------
Date:      11 May 1994 14:34:20 GMT
From:      bortz@cnam.cnam.fr (Stephane Bortzmeyer)
To:        comp.protocols.tcp-ip
Subject:   FTP problem between Europe and USA

We have a strange problem with ftp transfers.
The problem appeared two weeks ago. Before that, *all* ftp transfers 
went fine, either with France or with the USA, whatever the size of 
the files were. But now, we are unable to transfer some files from 
some US anonymous ftp servers.
The symptom is: not one byte comes to us, and we have a TCP reset a 
long time after.

Examples:

ftp://ftp.netcom.com/pub/livingston/sun/pm2_release_3.0.3.tar (the 
file is 1,4 Mbytes, the server is wu-ftpd "2.0WU(10)")

ftp://wuerlim.wustl.edu/pub/dicom/images/version3/mr2.tar the file 
is 5,6 Mbytes, the server is wu-ftpd "2.0WU(10)")

ftp://ftpboi.external.hp.com/pub/printers/laserjet/solaris22_23/hpnpC02.sol.tar 
(the file is 819 kbytes, the server is (HP-UX?) "Version 1.7.109.2")

ftp://ftp.eit.com/pub/web.software/getstats/getstats.gif (the file 
is 2 kbytes, the server is (?) "Version 1.1")

In each case, other files in the same directory can still be retrieved. 
For each of these servers, the problem is perfectly reproducible. It 
doesn't depend on the ftp client (plain Unix ftp, Macintosh's Fetch 
and Vax/VMS UCX-FTP have the same problem).

The problem is not CNAM-specific. Other sites in France have the same 
problem. I wrote to the ftp admins at ftp.netcom.com and they told 
me nobody else had the trouble. So I would like to ask for some tests 
from European sites. Can you retrieve the above files? 

From a network analyzer (MicroTechnology LANager) we saw the following 
exchange:

my-machine -> netcom : RETR the-file
netcom -> my-machine : SYN from ftp-data (20)                    
my-machine -> netcom : SYN ACK
netcom -> my-machine : ACK from ftp (21)
netcom -> my-machine : ACK from ftp-data (20)
netcom -> my-machine : "Opening BINARY..."
my-machine -> netcom : ACK 

and then nothing at least for one hour except sometimes a RST (full 
trace on demand).

Does any TCP/IP guru have an idea? 

Stephane Bortzmeyer           Conservatoire National des Arts et Metiers	
bortzmeyer@cnam.fr            Laboratoire d'Informatique
                              292, rue Saint-Martin			
tel: +33 (1) 40 27 27 31      75141 Paris Cedex 03
fax: +33 (1) 40 27 27 72      France	

"C'est la nuit qu'il est beau de croire a la lumiere." E. Rostand

http://web.cnam.fr/personnes/bortzmeyer/home_page.dom

					
	

-----------[000193][next][prev][last][first]----------------------------------------------------
Date:      Wed, 11 May 1994 14:56:45 GMT
From:      dox@bln.sel.alcatel.de (Guntram Dox)
To:        comp.protocols.tcp-ip
Subject:   where to find 'DHCP' -"tool to facilitate IPaddr renumbering" ?

In RFC1597 they mention a tool 'DHCP' on page 5, which facilitates renumbering
of IP addresses.

Where can I find more information on that ?

TIA,
regards Guntram.

PS: If you'd like to respond by email, please send TO: dox@fokus.gmd.de, but
followup preferred


 


-----------[000194][next][prev][last][first]----------------------------------------------------
Date:      Wed, 11 May 1994 14:57:04 GMT
From:      philm@hpqmditk.sqf.hp.com (Philip Melville)
To:        comp.protocols.tcp-ip,comp.dcom.sys.wellfleet,comp.dcom.sys.cisco
Subject:   Re: Variable IP subnets with RIP on Wellfleet

Arne Langmo (al@runit.sintef.no) wrote:

: We're on 5.77e, and get this mess.:
 
: I 05/11/94 13:16:53 ospf: 'Invalid IP mask, 193.219.32.0/255.255.224.0 ignored
: I 05/11/94 13:16:53 ospf: 'Invalid IP mask, 193.69.0.0/255.255.0.0 ignored'
: I 05/11/94 13:16:53 ospf: 'Invalid IP mask, 193.71.0.0/255.255.0.0 ignored'
: I 05/11/94 13:16:53 ospf: 'Invalid IP mask, 193.90.0.0/255.255.0.0 ignored'
                                                         ^^^^^^^^^^^
Class C addresses only allow the last 8 bits to be subnetted. So the first
3 octets in your subnet mask should be 255.255.255.
--
Philip Melville  Hewlett Packard South Queensferry Site, Scotland
Unix :   philm@hpsqf.sqf.hp.com
HPDESK:  Philip Melville / HPE600/IT
X400:  FN=Philip  SN=Melville  OU1=hpe600  ORG=hp  C=gb  ADMD=gold 400  PRMD=hp
or     FN=Philip  SN=Melville  OU1=hpe600  ORG=hp  C=us  ADMD=attmail   PRMD=hp

-----------[000195][next][prev][last][first]----------------------------------------------------
Date:      11 May 1994 15:46:51 GMT
From:      kline@tampico.cso.uiuc.edu (Charley Kline)
To:        comp.protocols.tcp-ip,comp.dcom.sys.wellfleet,comp.dcom.sys.cisco
Subject:   Re: Variable IP subnets with RIP on Wellfleet

>: I 05/11/94 13:16:53 ospf: 'Invalid IP mask, 193.219.32.0/255.255.224.0 ignored
>: I 05/11/94 13:16:53 ospf: 'Invalid IP mask, 193.69.0.0/255.255.0.0 ignored'
>: I 05/11/94 13:16:53 ospf: 'Invalid IP mask, 193.71.0.0/255.255.0.0 ignored'
>: I 05/11/94 13:16:53 ospf: 'Invalid IP mask, 193.90.0.0/255.255.0.0 ignored'
>                                                         ^^^^^^^^^^^
>Class C addresses only allow the last 8 bits to be subnetted. So the first
>3 octets in your subnet mask should be 255.255.255.

He's trying to supernet, ala CIDR. Wellfleet doesn't let you do that, but
under OSPF, variable length subnets seem to work fine. Haven't tested it
extensively though.

/cvk

-----------[000196][next][prev][last][first]----------------------------------------------------
Date:      11 May 1994 12:44:21 +0200
From:      mj@dfv.rwth-aachen.de (Martin Junius)
To:        comp.os.386bsd.misc,comp.os.linux.misc,comp.sources.wanted,comp.protocols.tcp-ip
Subject:   Source for yppasswdd?

Hi everyone!

Is there anywhere in the known universe of UNIX C source code an 
implementation of the NIS passwd server available? yppasswdd as used by 
yppasswd to set the password on the NIS server.

Is it maybe contained in the source of one of the various xxxBSD 
releases?

Please reply by mail. Thanx.

Martin
-- 
 _____ _____
|     |___  |   Martin Junius           FIDO:      2:2452/110.1
| | | |   | |   Communication Networks  Internet:  mj@dfv.rwth-aachen.de
|_|_|_|@work|   Aachen U of Technology  Phone:     ++49-241-8790220 (voice)

-----------[000197][next][prev][last][first]----------------------------------------------------
Date:      Wed, 11 May 1994 16:14:53 GMT
From:      sofiene@davinci (Kamoun Sofiene)
To:        comp.protocols.tcp-ip
Subject:   [Q]: RIP1 or RIP 2 ?



Hi there,

We are implementing IP routing in an 8-port Ethernet to FDDI bridge.
Should we implement RIP version 1 or RIP version 2, and why ?

Thanks for any help

sofiene kamoun
sofiene@auto.polymtl.ca


-----------[000198][next][prev][last][first]----------------------------------------------------
Date:      11 May 1994 17:09:45 GMT
From:      leonard@telcom.arizona.edu (Aaron Leonard)
To:        comp.protocols.tcp-ip
Subject:   Re: Dynamic IP addr & mail


In article <CpLC63.KM7@cc.umontreal.ca>, surprenc@JSP.UMontreal.CA (Surprenant Colin) writes:
|Hi,
|
|I would like to know what are the possibilities for the delivery of mail
|on systems that uses dialup accesses to the network + dynamic IP address 
|allocation. 
|
|One obvious solution would be to have a mail server on which users
|which use dialup IP log in to read/write mail. 
|
|Is there any other way?

I assume by "dynamic IP address allocation", you mean that a given
dialup client may get a different IP address each time it connects.

My suggestion, if this is what you're doing, is to stop doing this.
Instead, assign each dialup client THE SAME address each time it connects.
(We accomplish this using a Xylogics Annex as our comm server, but I'm
sure most comm servers can do the same.)  

Each client can be MXed by some mailhub.  When the client connects, that 
can trigger a dequeuing of the stored messages.

If you are stuck with "dynamic addresses", then a good solution would
be to have the clients run a POP or IMAP client ... this would avoid
the necessity of logging into your mail host.

Aaron

Aaron Leonard (AL104), <Leonard@Arizona.EDU>
University of Arizona Network Operations, Tucson AZ 85721
  \ Don't lock yourself into open systems. /

-----------[000199][next][prev][last][first]----------------------------------------------------
Date:      11 May 1994 17:54:28 GMT
From:      kannan@namao.cs.ualberta.ca (Kannan Thiruvengadam)
To:        comp.protocols.tcp-ip
Subject:   TCP Vs UDP

Hello,

I am developing an application (A Courseware) on
Ethernet. I intend to find out which performs better - TCP or UDP
for the purpose (which involves multicast of voice, full motion video
slow motion video and text.)

Comments beginning with "I don't think.." and "I have a feeling.." 
are most welcome. If you have experience of any kind (relevant to this work)  
please impart some of your knowledge to me. If you know of any 
reference on the material, thanks for pointing it out to me.

I will post details later. 

- Kannan

-----------[000200][next][prev][last][first]----------------------------------------------------
Date:      Wed, 11 May 1994 18:11:09 GMT
From:      robs@join.com (Rob Stevens)
To:        comp.unix.solaris,comp.protocols.tcp-ip
Subject:   SOCK_DGRAM bug in Solaris

Read from a SOCK_DGRAM socket in Solaris2 appears
to have a bug. I believe the behavior *should*
be as follows.

If you request a read of more bytes than are in the datagram
you get only what was in the datagram (that's the meaning
of a protocol which preserves record boundaries). If you
request a read of fewer bytes, you get what you ask for and
the rest is discarded.

Regardless of whether you agree with the last sentence,
that *is* the behavior in Solaris1 (specifically SunOS4.1.2).

On Solaris2.3 however, when you read fewer bytes, the remainder
of the datagram is not discarded. Your next read will pick up
where you left off.

Sockets in Solaris2 are implemented as streams, and consequently
support all generic streams ioctl's. I have tried using I_SRDOPT
to set the stream to message discard mode (RMSGD). Although
the ioctl returns success, this has no effect on the operation
of the socket. This isn't really surprising since the SOCKMOD
module is sitting atop the stream and presumably it can 
do whatever it likes in terms of returning data to the stream head.

Does anyone have any more information about this. Is SOCKMOD
documented anywhere? Is the behavior of Solaris2 generally
thought to be *correct* and that Solaris1 is at fault? Is
this on a list of *known* bugs in Solaris2?

Rob Stevens





-----------[000201][next][prev][last][first]----------------------------------------------------
Date:      Wed, 11 May 1994 19:03:56 GMT
From:      frank@tsh.com (Frank Mostek)
To:        comp.unix.aix,comp.protocols.tcp-ip
Subject:   Setting up a SLIP server

I need help setting up an RS6000/AIX as a SLIP server.  I know how to
use the slattach command in AIX, and I have successfully established a
SLIP connection from my PC at home to the RS6000.  I was only able to
see the RS6000, not any other nodes in the network.

I don't know how to let users dial in, login, and then type 'slip' and
have an IP address displayed and magically have the remote machine
become a part of our network.  I have seen this functionality with
Internet access providers.  Is this the proper thing to go?  What would
'slip' be, a shell script?

I currently only have one modem at my disposal, but I will get more
shortly.

Any help, documentation referals, etc... will be greatly appreciated.

Thanks.
-- 
Frank Mostek              (fmostek@tsh.com)
Technology Consultant     (708)390-6300
The Systems House, Inc    Fax: (708)803-4306
10500 Lunt Avenue         Rosemont,  IL 60018     

-----------[000202][next][prev][last][first]----------------------------------------------------
Date:      11 May 1994 20:07:03 GMT
From:      wdp@roadnet.ups.com (Bill Paris)
To:        comp.protocols.tcp-ip
Subject:   ICMP timestamp, optional?

Question about IP's Internet Control Message Protocol's (ICMP) 
Timestamp messages.  I came across the following statement in
Internet RFC 956 (D.L. Mills, 9/85):

    Support for the ICMP Timestamp messages is optional in the DoD
    Internet protocol suite, so it is not surprising that most hosts
    and gateways do not support it.

Is it really optional?  Wouldn't this mean that users of ICMP
timestamps, such as the BSD timed server, would be of limited use?
Where can I find the answer?

Thanks in advance,
Bill Paris

-----------[000203][next][prev][last][first]----------------------------------------------------
Date:      Wed, 11 May 1994 20:17:43 GMT
From:      veizades@ftp.com (John Veizades)
To:        comp.protocols.tcp-ip
Subject:   Re: where to find 'DHCP' -"tool to facilitate IPaddr renumbering" ?

In article <dox.768668205@slbh04> dox@bln.sel.alcatel.de (Guntram Dox) writes:
>From: dox@bln.sel.alcatel.de (Guntram Dox)
>Subject: where to find 'DHCP' -"tool to facilitate IPaddr renumbering" ?
>Keywords: DHCP
>Date: Wed, 11 May 1994 14:56:45 GMT
 
>In RFC1597 they mention a tool 'DHCP' on page 5, which facilitates renumbering
>of IP addresses.
 
>Where can I find more information on that ?

The DHCP specification is defined by RFCs 1541 aand 1542 implementations of 
the protocol are soon to be available from a number of vendors.  There is an 
article in the August 1993 Connexions journal.

Server implementations that I know of will be available shortly come from:

FTP Software Inc. for Windows
Microsoft with the Daytona Advance Server
Sun with their SolarNet product for Solaris

Client implementations are coming to most commercial TCP/IP implementations 
FTP Software's OnNEt product will contain clients for DOS and Windows.

John Veizades...
DHCP Implementation Manager and DHCP Contributor
FTP Software, Inc.

-----------[000204][next][prev][last][first]----------------------------------------------------
Date:      Wed, 11 May 1994 20:39:48 GMT
From:      landin@cherokee.nsuok.edu (Mark Landin)
To:        comp.protocols.tcp-ip,comp.sys.novell,bit.listserv.novell
Subject:   Subnetting and Netware

I need to break a Class C network into subnets, using each subnet in a 
locations which are connected basically as follows:

+----+    +------+                         +------+   +----+
| W1 |----|  S1  |-------------------------|  S2  |---| W2 |
+----+    +------+                         +------+   +----+

|            | |                             | |           |
+------------+ +-----------------------------+ +-----------+
 Network 1,             Network 2                 Network 1,
 Subnet A                                         Subnet B

111.111.111.2                                    111.111.111.254

W1 and W2 are PC's with an IP address in different subnets of Class C network
1. S1 and S2 are Netware servers which are connected to a backbone with
addresses in Class C Network 2. (The actual IP addresses are made up; I know
that .2 and .254 are not valid the way subnetting works.)

My question is whether another server on the backbone (call it S3) will be
able to figure out that to get to 111.111.111.2 you need to go through server
S1, but that 111.111.111.254 is accessible only through server S2.

I would try this, but the servers in question are inconveniently far apart,
and this looks like a pretty simple "yes-or-no" proposition. 

Thanks for any responses (that pertain to my question, that is!)


 
-- 
*-----------------------------------------------------------------------------*
*  Mark C. Landin					Northeastern St. Univ *
*  landin@cherokee.nsuok.edu					Tahlequah, OK *
*   "Living in the pools, they soon forget about the sea" - Neil Peart, RUSH  * 

-----------[000205][next][prev][last][first]----------------------------------------------------
Date:      Wed, 11 May 1994 21:46:18 GMT
From:      rstevens@noao.edu (W. Richard Stevens)
To:        comp.unix.solaris,comp.protocols.tcp-ip
Subject:   Re: SOCK_DGRAM bug in Solaris

> If you request a read of more bytes than are in the datagram
> you get only what was in the datagram (that's the meaning
> of a protocol which preserves record boundaries). If you
> request a read of fewer bytes, you get what you ask for and
> the rest is discarded.
> 
> Regardless of whether you agree with the last sentence,
> that *is* the behavior in Solaris1 (specifically SunOS4.1.2).
> 
> On Solaris2.3 however, when you read fewer bytes, the remainder
> of the datagram is not discarded. Your next read will pick up
> where you left off.

This "feature" is endemic to SVR4 and Solaris, and is IMHO a bug.
Check out the section "Datagram Truncation" on p. 160 of my recent
book "TCP/IP Illustrated" (Addison-Wesley, 1994).

> Is SOCKMOD documented anywhere?

You forgot to put a :-) after that.  :-)

	Rich Stevens

-----------[000206][next][prev][last][first]----------------------------------------------------
Date:      11 May 1994 22:28:31 GMT
From:      john@iastate.edu (John Hascall)
To:        comp.protocols.tcp-ip
Subject:   Re: BSD syslog(d), is there a RFC for this, is there an M$Windows port

Vonk C.J.S. <coert@athena.research.ptt.nl> wrote:
}Is the BSD syslogd protocol specified?  I went through the rfc-index.txt,
}but couldn't find it.  And is there a WinSock port of this deamon that
}I might run under Windows NT.  (yes, I know I might use the event deamon,
}but I'm developing for Open VMS).

   The word protocol is almost too strong a word.
   A syslog message is just ascii text.  By `convention'
   the form is:

       <###>Mmm dd hh:mm:ss hostname pid# programname: message text

   where ### is the a number representing the priority(severity)
   and possibly the facility.  When converted to numeric, the
   low 3 bits are the priority, above that the facility (see
   syslog.h on any unix box for values).

   By `convention', I mean that some syslogd's will add some
   of this info to the output if it doesn't "see" it -- and I
   suppose one could find syslogd's that wouldn't like it at all.

John
-- 
John Hascall                   ``An ill-chosen word is the fool's messenger.''
Systems Software Engineer
Project Vincent
Iowa State University Computation Center  +  Ames, IA  50011  +  515/294-9551

-----------[000207][next][prev][last][first]----------------------------------------------------
Date:      Wed, 11 May 1994 22:53:30 GMT
From:      andyg@austin.ibm.com (Andy Gillam)
To:        comp.unix.aix,comp.protocols.tcp-ip
Subject:   Re: Setting up a SLIP server

In article <CpnJML.rzJ@tsh.com> frank@tsh.com (Frank Mostek) writes:
>I don't know how to let users dial in, login, and then type 'slip' and
>have an IP address displayed and magically have the remote machine
>become a part of our network.  I have seen this functionality with
>Internet access providers.  Is this the proper thing to go?  What would
>'slip' be, a shell script?

Unfortunately, you're describing the sliplogin command.  It isn't available on
AIX 3.2.x.  It would be possible to implement this yourself as a shell
script without too much trouble.  The only real problem you're going to 
encounter is one of allowing your users to configure interface addresses.
This requires more than staff user permission.  Perhaps you could
write a program with setuid turned on which would do this for the user, but it
would be something of a security hole.

-Andy
-- 
DISCLAIMER:Any opinions reflected herein in no way reflect those of my employer.

-----------[000208][next][prev][last][first]----------------------------------------------------
Date:      11 May 1994 23:49:13 GMT
From:      zeeff@zip.eecs.umich.edu (Jon Zeeff)
To:        comp.protocols.tcp-ip
Subject:   encrypted telnet

Does anyone have a telnet/telnetd that encrypts the conversation?  Key
management is not really a concern for my application.


-----------[000209][next][prev][last][first]----------------------------------------------------
Date:      Thu, 12 May 1994 00:39:29 GMT
From:      al012@un.seqeb.gov.au ( ANTHONY LEE)
To:        comp.protocols.tcp-ip
Subject:   Using pipe with inetd


Dear all,

Recently I had to provide a service through a TCP/IP port via 
/etc/services and /etc/inetd.conf.  The service prints log
from the switching system of a power network.  I pipe the output
through grep to removed unwanted messages but for some reason
the output is buffered by grep such that nothing is printed until
the logging program has produced many many lines of logs.  I have
tried using setbuf from within grep (actually agrep, I modified the 
source) but it still the output is still being buffered.  I can
modify the logging program but a cleaner solution would be to use
grep.  Could someone please tell me how I can solve this piping problem.

Thank you

-- 
Anthony Lee				These are my opinions and not SEQEB.
SEQEB					
150 Charlotte Street			
Brisbane

-----------[000210][next][prev][last][first]----------------------------------------------------
Date:      Thu, 12 May 94 06:02:26 EDT
From:      ziegenfE@moravian.edu (Eric W. Ziegenfus)
To:        comp.protocols.tcp-ip
Subject:   Re: Programming with RPC. (PROBLEM)

In <1994May6.175301.15804@uab.es> root@news.uab.es (Gloria Hernandez. proyecto informatica) writes:

>Hello: 


>I am trying to understand the rpc Protocol Compiler  "rpcgen".
>But, I have a problem when I run an example application   (rprintmsg)
>The server program is running perfectly, but when i try to run the
>client program it displays the error message:


>		"Illegal instruction"


>when it try to execute the instruction
>   
>  		result = printmessage_1(cl,&message);
 
>(rprintmsg is the remote procedure)


>I have read about "RPC status code" but I dont Know what is the reason.

I just compiled that example up and had no problems. What environment are 
you working in? Did everything compile without errors?

ewz
-- 

Eric W. Ziegenfus        
ziegenfE@moravian.edu                                    No matter where you go
                              Ain't no time 2 hate.      there you are!
Te/vzemei un Bri/vi/bai          Grateful Dead              Buckaroo Bonzai


-----------[000211][next][prev][last][first]----------------------------------------------------
Date:      Thu, 12 May 1994 03:39:57 GMT
From:      billr@audi.optimation.co.nz (Bill Ryder)
To:        comp.os.ms-windows.networking.tcp-ip,comp.protocols.tcp-ip
Subject:   Has anyone seen this symptom with W4WG? (DANGEROUS ICMP behaviour)


IMPORTANT QUALIFICATION: This may be incorrect or apply to beta
                         revision software! Take it with as much salt
		         as you see fit! BUT - check out the symptom

SUMMARY:

  SYMPTOM: random network sessions , xterms, telnets etc failing for no
  obvious reason - they report "host unreachable" and die.

  CAUSE: W4WG sending "ICMP: Host Unreachable messages"

THE STORY:

This is a piece of mail I received from a friend in Sun Support. It
came from Massey University and details a VERY interesting problem
with Windows for Workgroups TCP/IP. Unfortunately I have no
information about what revision of W4WG they were using. I  had
nothing to do with the site or analysis of the problem so don't ask me
any questions about it. It is hearsay (hence not admissable in court)
from my perspective. However someone may be having these problems and
this may solve it for them.

THE MAIL:

> They found random network sessions, including database connections, xterms,
> telnets, etc were failing for no apparent reason.  They noticed that the
> xterms would be in a session, but report "host unreachable", and die.  
> Similar effects where noticed with the database sessions.
> 
> They started snooping the network for ICMP packets and found that a 
> "ICMP: host unreachable" had arrived shortly before the failure.  They
> then monitored those ICMP packets and found what was happening.
> 
> Recently, one department installed Windows for Workgroups with the Microsoft
> IP stacks.  The person at massey was not sure if this software was full 
> release yet.  The WFW machine would send out an ARP, the servers on the 
> network would respond, THEN the WFW machine would send a "ICMP: Host 
> Unreachable" to a random server with a random Hostname.  This server would
> then shutdown the connections to this random host.
> 
> Chopping the WFW machines of the Campus Network has removed a large number
> of strange network problems they were having.
> 
> Note that the WFW users had no problems at all, and never saw the failures.
 

-----------[000212][next][prev][last][first]----------------------------------------------------
Date:      Thu, 12 May 1994 03:42:55 GMT
From:      billr@audi.optimation.co.nz (Bill Ryder)
To:        comp.os.ms-windows.networking.tcp-ip,comp.protocols.tcp-ip
Subject:   Has anyone seen this symptom with W4WG? (DANGEROUS ICMP behaviour)


IMPORTANT QUALIFICATION: This may be incorrect or apply to beta
                         revision software! Take it with as much salt
		         as you see fit! BUT - check out the symptom

SUMMARY:

  SYMPTOM: random network sessions , xterms, telnets etc failing for no
  obvious reason - they report "host unreachable" and die.

  CAUSE: W4WG sending "ICMP: Host Unreachable messages"

THE STORY:

This is a piece of mail I received from a friend in Sun Support. It
came from Massey University and details a VERY interesting problem
with Windows for Workgroups TCP/IP. Unfortunately I have no
information about what revision of W4WG they were using. I  had
nothing to do with the site or analysis of the problem so don't ask me
any questions about it. It is hearsay (hence not admissable in court)
from my perspective. However someone may be having these problems and
this may solve it for them.

THE MAIL:

> They found random network sessions, including database connections, xterms,
> telnets, etc were failing for no apparent reason.  They noticed that the
> xterms would be in a session, but report "host unreachable", and die.  
> Similar effects where noticed with the database sessions.
> 
> They started snooping the network for ICMP packets and found that a 
> "ICMP: host unreachable" had arrived shortly before the failure.  They
> then monitored those ICMP packets and found what was happening.
> 
> Recently, one department installed Windows for Workgroups with the Microsoft
> IP stacks.  The person at massey was not sure if this software was full 
> release yet.  The WFW machine would send out an ARP, the servers on the 
> network would respond, THEN the WFW machine would send a "ICMP: Host 
> Unreachable" to a random server with a random Hostname.  This server would
> then shutdown the connections to this random host.
> 
> Chopping the WFW machines of the Campus Network has removed a large number
> of strange network problems they were having.
> 
> Note that the WFW users had no problems at all, and never saw the failures.


-----------[000213][next][prev][last][first]----------------------------------------------------
Date:      12 May 94 09:10:15
From:      zhao@crl.nmsu.edu (Z. Zhao)
To:        comp.protocols.tcp-ip.ibmpc,comp.protocols.tcp-ip,comp.protocols.ppp,comp.sys.sun.admin
Subject:   Good SLIP server for Sparc?

I have installed the SLIP server package coming with PC-NFS5.0 on a
Sparc10/4.1.3 and tried to launch some SLIP client applications, e.g.,
Trumpet-tcpman & Eudora-mail & WS_FTP, ..., for other people. I could
dial in the Sparc and enable SLIP from Trumpet-tcpman, but Eudora and
WS_ftp would say: connect failed. It seems they were truly negotiating
with the Sparc for while and failed to understand each other. 

The SLIP server setup in SUNOS needs different IP address for the same
machine and defines different network, hosts, ... from the hardwired
ether net using the same server. E.g., in the /etc/hosts

192.0.0.1	air	loghost	# for ethernet
192.0.255.1	air-gate	# for slipnet 

both represent the same machine. 

Is this the reason for Trumpet SLip client to fail connection?  Is
this a common setup in SLIP server software?

Has anyone made the psnfs-SLIP server working with Trumpet SLIP client
smoothly? If so, what is your experience? 

Is there any better SLIP server software for SUNOS and its setup is
more straight ?


I need your help, I need your answer, I need your experience.

Thanks in advance,


ZiZi

-----------[000214][next][prev][last][first]----------------------------------------------------
Date:      Thu, 12 May 1994 09:38:35
From:      milan@xyplex.com (Milan J. Merhar)
To:        comp.os.ms-windows.networking.tcp-ip,comp.protocols.tcp-ip
Subject:   Re: Has anyone seen this symptom with W4WG? (DANGEROUS ICMP behaviour)

In article <Cpo870.7JC@optimation.co.nz> billr@audi.optimation.co.nz (Bill Ryder) writes:
>  SYMPTOM: random network sessions , xterms, telnets etc failing for no
>  obvious reason - they report "host unreachable" and die.
>  CAUSE: W4WG sending "ICMP: Host Unreachable messages"

We have NOT seen activity of this kind coming out of Windows For Workgroups 
itself, but we HAVE seen gratuitous announcements being put onto the 
network by a DLL-based TCP/IP stack running under W4WG.

In our case, we suddenly lost TCP/IP connectivity in portions of our internal 
net connected via Routers. We determined that four or five network devices 
were SIMULTANEOUSLY sending phony OSPF route announcement messages,
disrupting the normal operation of the Routers.

Further investigation showed that the four or five devices were all PCs running
W4WG and Netmanage Chameleon V3.11. For some unknown reason, they suddenly
decided that whenever a (valid) OSPF route broadcast message was sent by a 
Router, they would turn it around and resend it as a unicast message with 
their own IP address as the source and the Router as the destination.

When Newt (the TCP/IP stack DLL) was unloaded (e.g.  by closing whatever 
application was using it), the PC stopped echoing OSPF messages.

Anyhow, I don't think that W4WG had anything to do with the problem. By the 
way, we also haven't seen the problem to date with Chameleon V4.0 thru V4.0.4.


Milan J. Merhar   milan@xyplex.com





















-----------[000215][next][prev][last][first]----------------------------------------------------
Date:      Thu, 12 May 1994 04:51:17 GMT
From:      ueen@pool.info.sunyit.edu (Eugene E. Ninestein)
To:        comp.protocols.tcp-ip
Subject:   Re: TCP/IP   D.E. Comer Volume III

In article <johnston.40.00061C34@hookup.net> johnston@hookup.net (Stewart Johnston) writes:
>TCP/IP   D.E. Comer Volume III
>
>I have just purchased "Internetworking with TCP/IP volume III" by Douglas E. Comer and David
>L. Stevens, published by Pentice Hall.  On the back of the book there is a note that reads:
>	"  - Includes programs available from the publisher and by FTP. "
>
>I assume this means the example code in this book is available through FTP.  Unfortunately the
>authors failed to publish the FTP sight address.
>
>Could someone please help me find the FTP address for the above source code.
>
>Your help would be greatly appreciated.
>
>Thanks
>Stewart Johnston
>johnston@hookup.net

Stewart,
	It's at arthur.cs.purdue.edu in /pub/dls and the file is called
v3.tli.dist.tar.Z  :)

Gene
-- 
Gene Ninestein                                     Dart Communications      
Product Development        Phone: 315.841.8106     TCP/IP Tools and Services 
ueen@sunyit.edu (for now)  Fax:   315.841.8107     #include <std_disclaimer>

-----------[000216][next][prev][last][first]----------------------------------------------------
Date:      Thu, 12 May 1994 05:09:02 GMT
From:      daford@netcom.com (David Ford)
To:        comp.protocols.tcp-ip
Subject:   Mac to PC TCP-IP link

A friend of mine has a Mac and PC and wants to connect them together 
without going apple-talk like the Coactive Connector.  I was thinking I 
could put the two together via ethernet.  I was thinking about picking up 
a tcp-ip adon for both and go.  Any suggestions?  Also, what ftp sites 
can I find things at to do this?  thanks.


-- 
                                             daford@netcom.com

-----------[000217][next][prev][last][first]----------------------------------------------------
Date:      Thu, 12 May 1994 05:28:14 GMT
From:      paul@atlas.abccomp.oz.au (Paul Brooks)
To:        comp.protocols.tcp-ip,comp.dcom.sys.cisco
Subject:   Re: multiple gateways

In article <2q66d4$evl@agate.berkeley.edu> sean@oak.his.ucsf.edu writes:
|Situation: multiple LANs interconnected via cisco routers and T1s.
|
|Problem:  most hosts (PCs, Macs, terminal servers) only know about one
|'gateway' off the local net.  If a router dies the hosts don't know
|about another one on the same LAN.  From the end user perspective there
|is no redundancy.
|
|Solutions?

Bug your PC, Mac and terminal server TCP/IP providers to implement 
RFC 1256 ICMP Router Discovery.

-- 
Paul Brooks              |paul@abccomp.oz.au       |Emerging Standard:
TurboSoft Pty Ltd        |pwb@newt.phys.unsw.edu.au|  one that has not yet
579 Harris St., Ultimo   |                         |  been superseded.
Sydney Australia 2007    |ph: +61 2 281 3155       |  

-----------[000218][next][prev][last][first]----------------------------------------------------
Date:      Thu, 12 May 94 12:51:04 PDT
From:      cclarke%ghs.uucp@usc.edu (Chris Clarke)
To:        comp.protocols.tcp-ip
Subject:   New SCO UNIX User has TCP/IP/NFS/PPP Questions


               New SCO UNIX User has TCP/IP/NFS/PPP Questions
               ----------------------------------------------

Are there any public domain implementations of TCP/IP/NFS/PPP for UNIX out 
there?

Have any been ported to SCO UNIX?

All the best.

Chris Clarke
cclarke%ghs.uucp@usc.edu
--------------------------------------------------------------------------------

-----------[000219][next][prev][last][first]----------------------------------------------------
Date:      Thu, 12 May 1994 07:19:09 GMT
From:      dirk@incom.de (Dirk Koeppen)
To:        comp.protocols.tcp-ip,comp.protocols.tcp-ip.ibmpc
Subject:   Discussion: Larger BOOTP vendor field

Dear BOOTP users,

we are doing BOOTP/TFTP ROMs for PC's for a long time now and many customers  
complain that the BOOTP vendor field is too small. Many customers use the  
vendor field to include default login names or mount directories.

The first step was to implement the extended filename to transfer the vendor  
information that does not fit into the 64 bytes. This helps but if you have a  
network with 10 or 50 concurrent booting PC's, your server will not be able to  
fork all the TFTP daemons in an acceptable time.

So I would like to put the following into discussion:

Why don't we make an add-on to BOOTP as done in DHCP: If the client sets the  
option 57 (DHCP maximum message size) to a value, the server sends a reply of  
that maximum size to the client. The vendor field in a BOOTP request may  
therefore look like this:

	- RFC-1048 magic cookie (4 bytes) 
	- Tag number 57 set to maximum package size (4 bytes)
	- Tag number 255 (end of vendor field, 1 byte)

I see the following advantages:

	- Request complies to existing RFCs.
	- Older BOOTP daemons will not be affected as the request is still 
	  the old RFC size.
	- Older BOOTP forwarders or relays are still able to handle the
	  smaller BOOTP request, the reply may be sent directly to the
	  client in order to bypass older BOOTP forwarders.

And the following disadvantages:

	- When the BOOTP reply is broadcasted, older clients may receive a 
	  larger reply. Anyway they should be able to handle this as the
	  reply is not intended for them.
	- If the BOOTP reply is forwarded through a BOOTP gateway, the
	  gateway will not be able to handle the larger package size. 
	  

Please let me know your suggestions on extending the BOOTP RFC !
As I often loose news postings, please send me a copy of your posting via email  
to dirk@incom.de.

Thanks,
dirk

---
                   Dirk Koeppen EDV-Beratungs-GmbH
            Holzwiesenweg 22 * D-63073 Offenbach * Germany
  Phone: +49 69 89 3000 * FAX: +49 69 89 3004 * email: dirk@incom.de
	 


-----------[000220][next][prev][last][first]----------------------------------------------------
Date:      12 May 1994 08:31:14 GMT
From:      heathh@wrath.ugcs.caltech.edu (Heath I Hunnicutt)
To:        comp.os.ms-windows.networking.tcp-ip,comp.protocols.tcp-ip
Subject:   Re: Has anyone seen this symptom with W4WG? (DANGEROUS ICMP behaviour)

The previous post is complete and total bunk.  

- I am on the network at a campus with a LARGE (>1000) number of UNIX hosts
using TCP/IP, and many (>50) machines on the same net running WFWg with 
TCP/IP stacks and NetBEUI.  Some of these machines are running pre-release
Daytona TCP/IP stacks, to boot.  Not once have I ever heard a single tale
of such a situation occurring here.  And I would, as I am sort of a Windows
networking guru around here.  

- Consider the source of your admittedly heresay report: Sun Select.  Would
Sun have anything to lose if MS' networking products captured the market?
You bet they would.  It's not hard to imagine why they wold make up stories
like the one you report.

- The last bit is that the report claims that WFWg "randomly" chooses the 
host to challenge.  It must be getting awfully lucky to actually hurt 
machines at the site named.  Consider that there are 2^32 possible IP
addresses to hit, and perhaps only 2^10 or 2^12 machines actually on their
local network.

- Finally, the fact that Sun's TCP/IP stacks allows ICMP packets to disrupt
established connections, in non-conformance with the Hosts Requirements RFC.
Surely, this must be somewhat embarrassing to Sun Select, eh?

Later,
Heath


-----------[000221][next][prev][last][first]----------------------------------------------------
Date:      12 May 1994 10:27:39 GMT
From:      droms@regulus.cs.bucknell.edu (Ralph E. Droms)
To:        comp.protocols.tcp-ip
Subject:   Re: where to find 'DHCP' -"tool to facilitate IPaddr renumbering" ?

In article <dox.768668205@slbh04> dox@bln.sel.alcatel.de (Guntram Dox) writes:

   From: dox@bln.sel.alcatel.de (Guntram Dox)
   Keywords: DHCP
   Date: Wed, 11 May 1994 14:56:45 GMT

   In RFC1597 they mention a tool 'DHCP' on page 5, which facilitates renumbering
   of IP addresses.

   Where can I find more information on that ?

   TIA,
   regards Guntram.

   PS: If you'd like to respond by email, please send TO: dox@fokus.gmd.de, but
   followup preferred

Check out RFCs 1541, 1533 and 1534.  Also, there's the mailing list
host-conf@sol.cs.bucknell.edu, which is focused on DHCP.

--
- Ralph Droms                 Co-Director, Computer and Communication Services
  droms@bucknell.edu          Associate Professor of Computer Science
  (717) 524-1795              Bucknell University
  (717) 524-1790 (fax)        Lewisburg, PA 17837

-----------[000222][next][prev][last][first]----------------------------------------------------
Date:      12 May 1994 22:51:28 -0700
From:      makey@VisiCom.COM (Jeff Makey)
To:        comp.sys.sun.admin,comp.protocols.tcp-ip,comp.dcom.lans.ethernet
Subject:   Two IP networks on one ethernet = broadcast storms

To ease the transition of several dozen hosts to new IP numbers, I
want to support both the new and the old network numbers on the same
ethernet wire.  To route IP packets between the old and the new, I
have a SPARCstation 1 running SunOS 4.1.1 with 2 ethernet interfaces.
It seems to work fine, except that a broadcast IP packet on the wire
will turn into a broadcast storm.

When broadcast packets for network X.X.X.0 go onto the the ethernet
with the MAC broadcast address ff:ff:ff:ff:ff:ff, the Sun's other
ethernet interface (configured for network Y.Y.Y.0) picks up the
broadcast packet as it should, but pays no attention to the fact that
it is not addressed to that IP interface.  The Sun then seems to
think, "I know the route to X.X.X.255," and promptly dumps the packet
out the X.X.X.0 interface (onto the same wire).  This goes on until
the packet's TTL expires.

I am looking for suggestions on how to turn off IP forwarding only for
broadcast packets.  Most (perhaps all) of my broadcast storms consist
of RIP packets, and during this transition time I really don't want to
have to turn off routed and use static routing.  (There are other
routers on the network that broadcast RIP packets, too.)  Maybe there
is a SunOS patch?

                           :: Jeff Makey
                              makey@VisiCom.COM

           Disclaimer: All opinions belong only to their author.
Department of Tautological Pleonasms and Superfluous Redundancies Department

-----------[000223][next][prev][last][first]----------------------------------------------------
Date:      Thu, 12 May 1994 12:55:53 GMT
From:      dowd@acsu.buffalo.edu (Patrick Dowd)
To:        comp.protocols.tcp-ip
Subject:   SIGCOMM'94 Advance Program




                           Advance Programme
                      ACM SIGCOMM'94 CONFERENCE
       Communications Architectures, Protocols and Applications

                      University College London
                              London, UK

                    August 31 to September 2, 1994
                 (Tutorials and Workshop, August 29-30)


                             Sponsored by
         The ACM Special Interest Group of Data Communication

This  conference provides an international  forum for the presentation
and discussion of communication network applications and technologies,
architectures,  protocols,  algorithms, and  performance models.   The
conference  and tutorials will be conducted on  the University College
London, London England.


		  ----------------------------------
		  T E C H N I C A L    P R O G R A M
		  ----------------------------------

Monday 29 August 1994

*  7:30AM - 5:00PM 
   Tutorial and Conference Registration
   UCL CS Department, Pearson Building

*  9:00AM - 5:00PM, Tutorial T1
   "Personal Communication Services and Networks"
   Zygmunt Haas (AT&T Bell Labs)
   UCL CS Department, Pearson Building

*  9:00AM - 5:00PM, Tutorial T2
   "Protocol Performance"
   David D. Clark (MIT)
   UCL CS Department, Pearson Building

Tuesday 30 August 1994

*  7:30AM to 5:00PM 
   Tutorial and Conference Registration
   Edward Lewis Lecture Theatre, Windeyer Building

*  9:00AM - 5:00PM, Workshop W1
   "Topics    in   High   Performance   Networking   Support   of
    Distributed Systems" 
   Derek McAuley (University of Cambridge)
   UCL CS Department, Pearson Building

*  9:00AM - 5:00PM, Tutorial T3
   "Fiber Optic Networks"
   Paul  E. Green, Jr. (IBM Corporation)
   UCL CS Department, Pearson Building

*  9:00AM - 5:00PM, Tutorial T4
   "Multimedia Conferencing on the Internet"
   Van Jacobson (Lawrence Berkeley Laboratories)
   Edward Lewis Lecture Theatre, Windeyer Building

*  9:00AM - 5:00PM, Tutorial T5
   "Asynchronous Transfer Mode"
   Rainer Handel (Siemens Munich)
   UCL CS Department, Pearson Building

*  5:30PM - 8:30PM
   Welcoming Reception
   The Quad at University College London


Wednesday 31 August 1994

*  7:30AM to 5:00PM 
   Conference Registration
   Edward Lewis Lecture Theatre, Windeyer Building

*  9:00AM - 10:00AM
   Session 1: Keynote Address
   (1994 ACM SIGCOMM Award Winner)
   Edward Lewis Lecture Theatre, Windeyer Building

*  10:30AM-12:30PM
   Session 2: Protocol Performance
   Experiences with a High-Speed Network 

   Adaptor: A Software Perspective (Best Student Paper)
   P. Druschel (University of Arizona), L.L. Peterson (University of 
   Arizona), & B.S. Davie (Bellcore) 

   User-space Protocols Deliver  High Performance to Applications on a
   Low-Cost Gb/s LAN 
   A. Edwards, G. Watson, J. Lumley, D. Banks, 
   C. Calamvokis, & C. Dalton (Hewlett-Packard Labs, Bristol)  

   TCP Vegas: New Techniques for Congestion Detection and Avoidance
   L.S. Brakmo, L.L. Peterson, & S.W. O'Malley (University of Arizona)

   A Structured TCP in Standard ML
   E. Biagioni (Carnegie Mellon University)

*  12:30PM  - 2:00PM
   Lunch

*  2:00PM-3:30PM
   Session 3: Congestion Management

   Making Greed Work  in Networks: A  Game-Theoretic  Analysis of
   Switch Service Disciplines 
   S. Shenker (Xerox PARC)

   Scalable Feedback Control for Multicast Video Distribution in the Internet
   J. Bolot (INRIA), T. Turletti (INRIA) & I. Wakeman 
   (University College, London)

   Statistical Analysis of Generalized Processor 
   Sharing Scheduling Discipline
   Z.-L. Zhang, D. Towsley, & J. Kurose (University of Massachusetts)

*  4:00PM-5:30PM
   Session 4: ATM Flow Control

   The Dynamics of TCP Traffic over ATM Networks
   A. Romanow (Sun Microsystems) & S. Floyd (Lawrence Berkeley Labs)

   Reliable and Efficient Hop-by-Hop Flow Control
   C. Ozveren (DEC, Littleton), R. Simcoe (DEC, Littleton)  & 
   G. Varghese (Washington University, St. Louis)

   Credit Update Protocol for Flow-Controlled ATM 
   Networks: Statistical Multiplexing and Adaptive Credit Allocation
   H.T. Kung (Harvard University), T.  Blackwell (Harvard 
   University), & A. Chapman (BNR)

*  7:30PM - 10:00PM
   SIGCOMM Social: Reception and Dinner 
   The Dinosaur Room, Natural History Museum 
   (Tickets  Required)


Thursday 1 September 1994

*  7:30AM to 5:00PM 
   Conference Registration
   Edward Lewis Lecture Theatre, Windeyer Building

*  8:30AM - 10:00 AM
   Session 5: Internet Routing

   Flexible Routing and Addressing for a Next Generation IP
   P. Francis (NTT Software Labs) & R. Govindan (Bellcore)

   An Architecture for Wide-Area Multicast Routing
   S. Deering(Xerox PARC), D. Estrin (University of Southern 
   California), D. Farinacci (Cisco Systems), V. Jacobson 
   (Lawrence  Berkeley Labs), C.-G.  Liu (University of  Southern
   California) & L. Wei (University of Southern California) 

   Distributed Routing Based on Link-State Vectors
   J. Behrens & J.J.  Garcia-Luna-Aceves (University of 
   California at Santa Cruz)

*  10:30AM-12:00PM
   Session 6: ATM Switching and Signalling

   Signaling and Operating System Support for  
   Native-Mode ATM Applications
   R. Sharma & S. Keshav (AT&T Bell Labs)

   Experiences of Building ATM Switches for the Local Area
   D.R. McAuley, R.J. Black & I.M. Leslie (University of Cambridge)

   Controlling Alternate Routing in General-Mesh 
   Packet Flow Networks
   S. Sibal (RPI) & A. DeSimone (AT&T Bell Labs)

*  12:00PM  - 1:30PM
   Lunch

*  1:30PM-3:00PM
   Session 7: Nueral and Optical Networks

   On Optimization of Polling Policy Represented 
   by Neural Network
   Y. Matumoto (I.T.S., Inc., Japan)

   An Optical Deflection Network
   J.  Feehrer  (University of  Colorado,  Boulder),  L.  Ramfelt
   (University of Colorado, Boulder/Royal Institute of Technology,   
   Stockholm), & J. Sauer (University of Colorado, Boulder) 

   Conflict-Free Channel Assignment for an Optical 
   Cluster-Based Shuffle Network Configuration
   K.A. Aly (University of Central Florida)

*  3:30PM-5:30PM
   Session 8: Selected Topics

   MACAW: A Media Access Protocol for Wireless LANs
   V. Bharghavan (UC Berkeley), A. Demers (Xerox PARC), 
   S. Shenker (Xerox PARC) & L. Zhang (Xerox PARC)

   Asymptotic Resource Consumption in Multicast 
   Reservation Styles
   D.J. Mitzel (University of  Southern  California) & S. Shenker
   (Xerox PARC) 

   Highly Dynamic  Destination-Sequenced  Distance-
   Vector Routing  (DSDV) for Mobile Computers
   C.E. Perkins & P. Bhagwat (IBM, Watson Research Center)

   A Methodology for Designing Communication Protocols
   G. Singh (Kansas State University)

*  5:30PM - 6:30PM
   SIGCOMM Business Meeting


Friday 2 September 1994

*  8:30AM - 10:00AM
   Session 9: Traffic Models

   Wide-Area Traffic: The Failure of Poisson Modeling
   V. Paxson & S. Floyd (Lawrence Berkeley Labs)

   Analysis, Modeling and Generation of Self-Similar 
   VBR Video Traffic
   M.W. Garrett & W. Willinger (Bellcore)

   An Algorithm for Lossless Smoothing of MPEG Video
   S.S. Lam, S. Chow, & D. Yau (University of Texas, Austin)

*  10:30AM-12:00PM
   Session 10: Host Software

   USC: A Universal Stub Compiler
   S.W. O'Malley, T. Proebsting, & A. Montz (University of Arizona)

   An Object-based Approach to Protocol Software Implementation
   C.-S. Liu (Chung Yuan Christian University, Taiwan)

   Improved Algorithms for Synchronizing Computer Network Clocks
   D.L. Mills (University of Delaware)

*  12:00PM - 12:15PM
   Closing Session
   
Note: Program subject to change.
				   
			  -----------------
			  T U T O R I A L S
			  -----------------
				                       
Tutorial T1 
-----------
Zygmunt Haas, AT&T Bell Labs
"Personal Communication Services and Networks"

The  recent explosion  of interest  in  wireless and mobile  networks,
stimulated  by  the effort  of  Personal  Communication  Services  and
Networks (PCS & PCN) to  be  deployed at  the  beginning  of  the next
century,  suggests   the  enormous   technological,   scientific,  and
commercial potential in this field. The subject of wireless and mobile
communication  integrates  the  large body  of  knowledge  accumulated
through   the  traditional  radio   research,  the   large  networking
experience accumulated through the proliferation of LANs and WANs, and
the  vision  of  ubiquitous  connectivity anywhere,  at anytime,  with
anyone, and in any format.
   The  tutorial exposes both the theoretical  and  the practical
aspects  of mobile  networking,  from  a  networking  and  application
perspective.   We   will   present   the  concept,  architecture,  and
functionality of Personal Communications Services  and Networks (PCS &
PCN)  and  Universal  Personal  Telecommunications (UPT)  and  we will
describe the most  common  platform  for  mobile  communications:  the
wireless systems. In  particular, systems such  as cellular, cordless,
and satellite will be  discussed. Existing  and in-progress  standards
are also outlined.
   Finally, an abundance of examples of  the  wireless and mobile
networks will be described, giving realism to the presented material. 
TOPICS:
* Elements of Wireless Mobile Communications
* Wireless Services and Applications
* The Cellular Concept 
* The Cordless Concept 
* Digital Communication Networks
* Local-Area Wireless Data Access
* Wide-area Wireless Data Access
* Mobile Satellite Communications
* Standardization of Wireless Networks
* PCS/PCN and UPT
* Summary: Where we have started and where are going .

Zygmunt Haas received his B.Sc. in EE in 1979 and M.Sc. in EE in 1985,
both with  Summa Cum Laude.   From  1979  till 1985  he worked for the
Government of Israel.  In  1988,  he  earned  his Ph.D.  from Stanford
University researching fast packet-switched networks, and subsequently
joined AT&T Bell Laboratories in Holmdel NJ, where he is now  a Member
of  Technical  Staff in the Wireless Networks Department.  Dr. Haas is
an author of  numerous  technical papers and holds several  patents in
the field  of high-speed networking, wireless  networks,  and  optical
switching.  He has organized  several workshops and  served as a guest
editor for  JSAC issues.  Dr. Haas is a Senior  Member of IEEE and his
interests  include:   mobile   and  wireless  communication  networks,
personal  communication services,  high-speed communication protocols,
and optical switching.



Tutorial T2 
-----------
David D. Clark, MIT
"Protocol Performance"

Getting proper  performance from  a  network  or  protocol  is often a
difficult task. This tutorial uses examples from the Internet (TCP/IP)
protocol suite to illustrate critical performance issues. The focus is
on  providing  real-world  advice  on  how  to  design  and  implement
protocols in ways that  avoid performance  problems. The  presentation
will include  examples  of various  performance  problems and  how  to
detect and recognize them.

Topics
* Performance issues (reliability, throughput and delay)
* Implementation bottlenecks
* Specifications and their limitations
* Heterogeneity and its impact on implementation
* Network dynamics
* Visualizing protocol performance
* Limits of protocol performance

Dr. David Clark  is a senior research scientist  at MIT Laboratory for
Computer Science  and  a  recipient  of the ACM SIGCOMM  Award. He has
worked  on TCP/IP  since the  mid-1970s and  from  1981  to  1989  was
chairman of the Internet Activities Board. He is widely known for  his
insight into  protocol design and performance  and for  his  skill  in
identifying and  eliminating  myths about  protocol implementation and
performance.  His current areas  of  research include high-performance
networks,   the  evolution   of  the  Internet,  ATM  and  information
networking. He received his doctorate from MIT in 1973.

Tutorial T3 
-----------
Paul E. Green, Jr., IBM
"Fiber Optic Networks"

Fiber  optic   technology  has  completely  transformed  the  internal
operation of the world's telephone networks and is beginning to impact
local  computer networks.  Compared to  the  voice  grade  phone  line
technology, which defined  most of the  network architectures  that we
are  still  living with  today,  fiber  offers ten orders of magnitude
better bandwidth and an  equal  improvement in  achievable  bit  error
rate.  By use of WDM and circuit switching, the additional benefits of
protocol transparency can be achieved.
   There is a widespread  feeling that  the generation of network
that  will  follow today's ATM and upgraded  Internet structures might
very  well   be  based   on  techniques  that  directly  unlock   this
revolutionary improvement at the physical level.
   The  course is  devoted  to  the  new  class  of "all-optical"
networks  that attempt  to  do  this.  The  lecturer  will  cover  the
optoelectronic components  involved and will  also treat  some  of the
network  architectural  consequences,  the  regulatory   and  economic
picture, and review some systems already implemented.

TOPICS 
* Motivating fiber optic networks
* Fibers, couplers and taps
* Optical resonant structures
* Laser diodes and amplifiers
* Optical receivers
* System considerations
* Network topologies and link budgets
* Protocols, layers and network control
* Some implemented systems
* Status and prospects

Paul E. Green, Jr, is Manager of Advanced Optical Networking Member at
IBM Research in Hawthorne, NY.  He received the ScD degree from M.I.T.
in  1953, and after some  years at M.I.T. Lincoln Laboratory, where he
made pioneering  contributions to spread spectrum, adaptive receivers,
radar astronomy and seismic data processing, he joined IBM Research in
1969.  At  IBM  he has held  a  variety  of  management and  Corporate
Technical  Staff  positions. His technical interests  have centered on
computer  network  architecture,  and  he  has  received  several  IBM
Outstanding Innovation Awards for his  role in the initial formulation
and promotion of Advanced Peer to  Peer Networking,  now the basis for
further evolution of IBM's System Network Architecture. He is a member
of   the  National  Academy  of   Engineering,  in  1983   was   named
Distinguished Engineering Alumnus by North Carolina  State University,
and received the IEEE's Simon Ramo  Medal in 1991. He is the author of
many  technical   papers,  has   edited   several  books  on  computer
communications,  and  is  the  author  of  the  textbook  Fiber  Optic
Networks,  published  by  Prentice  Hall  in  June,1992.  He  has been
President of  both the IEEE  Communication  Information Theory Society
and the Communication Society.

Tutorial T4 
-----------
Van Jacobson, LBL
"Multimedia Conferencing on the Internet" 

An architectural overview  and detailed walk-through of  the protocols
and applications that provide  real-time, multiparty, audio, video and
shared workspace conferencing on today's Internet.
   Experiments and demonstrations over the Internet MBONE and the
DARTNET  testbed  have  shown   that   multimedia   and   conferencing
applications can indeed work  over  IP internets.  Playback algorithms
that  adapt  to  variations  in  network  delay  (such  as   VAT)  and
information  distribution  algorithms  designed  to  facilitate shared
workspaces (such  as those used in the  shared  whiteboard)  have made
these sorts  of applications  possible. This tutorial describes  these
algorithms and the applications that use them.
Topics
* IP as a real-time infrastructure: multicasting and queueing
* Adaptive Playback: VAT
* Managing Sessions: SD
* Managing Shared Workspaces: Shared Whiteboard
* Implications for the future of IP

Van Jacobson is a senior researcher at Lawrence Berkeley Laboratories,
where he works on real-time system performance, protocol and operating
system performance.  He is widely known for his groundbreaking work on
TCP/IP  performance,  TCP/IP   congestion  control,  and  support  for
multimedia  applications  on the  Internet. He is  the recipient of  a
number  of  awards and  teaches  periodically  at  U.C.  Berkeley  and
Stanford University.  

Tutorial T5 
-----------
Rainer Handel, Siemens Munich
"Asynchronous Transfer Mode" 

The  tutorial  will  provide  a   comprehensive  introduction  to  the
Asynchronous Transfer Mode (ATM). Both technical and marketing aspects
of ATM will be addressed. ATM specification is not yet complete but in
a state that allows implementations which are basically compliant with
a worldwide agreed, unique standard supporting  data, voice, image and
multimedia applications.
       The presentation of the concept of ATM networking  will include
the ATM  protocol reference model,  the architecture of  ATM networks,
interfaces  and procotols, traffic  control  and resource  management,
signalling, operational  aspects,  ATM evolution  and  internetworking
aspects, and of course a detailed description of the ATM layer and ATM
adaptation layer  functions. An overview of how ATM cells are switched
and  transmitted  will  also  be  given. The  possible use of ATM in a
business   and  residential  environment  and  its  market  acceptance
depending on product availability, cost  and feature offerings will be
clarified.   
TOPICS:  
* High speed networks  
* ATM concept 
* ATM protocols  
* ATM interfaces 
* interworking and evolvability
* market  aspects 
* switching and transmission products 
* network  implementations and  service offerings  

Rainer Handel  has  been with Siemens  (Public Communications Networks
Group)  since  1978 doing system  design and software  development for
switching  systems,  ATM  conceptual  and  standardization  work,  ATM
network and  product planning,  and currently long-term telecom market
and  technology trend evaluation. For several years  he  was active in
the standards bodies  CCITT, ETSI and T1, and is the author of several
papers and a book on ATM.

Workshop W1 
----------
Derek McAuley, University of Cambridge
"Topics in High Performance Networking Support of Distributed Systems" 

This  one day workshop will present the experiences of the speakers in
building various  components  of  distributed  systems  which  aim  to
effectively  utilise modern  high performance  networks. This workshop
consists of 4 talks. Each  talk will be 60 minutes with 15 minutes for
discussion.

1. The CHORUS Communication Architecture, Marc Rozier

The   communication   service  is   a  key  component  of  the  CHORUS
micro-kernel  architecture. First,  it provides  the  basic  framework
allowing  efficient   modular  operating  system  implementations.  By
dramatically reducing the overhead of  local communications, it is key
to the success of such  serverized implementations, which are now able
to  compete  with  monolithic  implementations.  Second,  it  provides
efficient, network-transparent, communication  services, well  adapted
to the distribution of  the operating system servers.   In particular,
it makes  possible  the  implementation  of UNIX systems  on massively
parallel architectures, offering a single system image to their users.
This tutorial will  address the  various aspects of this communication
architecture, from  the definition of the communication  services,  to
some  aspects of  its  implementation.  Emphasis  will  be  placed  on
insights from previous versions of this service.

2. The Organization of Networks in Plan 9, Rob Pike

In  a distributed system networks  are  of paramount importance.  This
tutorial   describes  the   implementation,   design  philosophy   and
organization of  network support  in  Plan  9. Topics include  network
requirements  for  distributed  systems,  our  kernel  implementation,
network naming, user interfaces and performance. We also observe  that
much of this organization is relevant to current systems.

3. Mixed media applications, David Tennenhouse

WWW  is a rapidly growing  phenomena which highlights the  interesting
applications possible with mixed media types. From experience with the
WWW this tutorial will  address the issues  raised in supporting these
mixed media types and the problems  in building systems  which support
media with time constraints.

4. What can you do with ATM today?, Derek McAuley

ATM must now be officially a  bandwagon. Some will tell you it  solves
all the world's problems because it was designed to, while others will
say  it's  good  for  nothing.  The  reality  and  hype  are  hard  to
distinguish. This talk will address what ATM can be used for today and
highlight those features for which it  is rightly criticised not least
of  which is  end-system  integration.  The talk could  be  subtitled,
"Difficult questions to ask your ATM salesman''.


Marc Rozier is the head  of the Micro-Kernel Department  within Chorus
systemes. He graduated from Ecole Nationale Superieure Informatique et
de  Mathematiques Applique'es  de Grenoble (ENSIMAG) before earning  a
doctor's   degree   in  Computer   Science  from   Institut   National
Polytechnique  de Grenoble (INPG). In 1981-82,  he was involved in the
CESAR  project at  IMAG  (Grenoble),  working  on  the  Validation  of
Distributed Systems. He gained experience in programming languages for
distributed applications  and distributed  systems. He joined INRIA in
1982 as  a  researcher  in  the CHORUS  distributed  operating  system
project. In 1987, he became one of the founders of Chorus systemes. He
is one of the main designers of the CHORUS-v3 Micro-Kernel technology.
He is the author of several publications in international journals and
conferences.

Rob Pike is  well known for his appearances on "Late Night with  David
Letterman",  is  also  a  Member  of  Technical  Staff  at  AT&T  Bell
Laboratories in Murray Hill, New Jersey, where he has been since 1980,
the same year he won  the  Olympic silver medal in Archery. In 1981 he
wrote the first  bitmap window system  for Unix systems, and has since
written ten  more.  With Bart Locanthi he designed  the Blit terminal;
with Brian Kernighan he  wrote The Unix  Program- ming Environment.  A
shuttle mission nearly  launched a gamma-ray telescope he designed. He
is a Canadian citizen and has never written a program that uses cursor
addressing.

David Tennenhouse is  an Assistant  Professor of Computer Science  and
Electrical Engineering at MIT's Laboratory for Computer Science. He is
leader  of  the  Telemedia,  Networks  and  Systems  Group,  which  is
addressing  systems  issues  arising  at   the  confluence   of  three
intertwined  technologies: broadband networks,  high  definition video
and distributed computing.   David  studied electrical  engineering at
the  University of Toronto, where he received  his B.A.Sc. and M.A.Sc.
degrees.  In 1989 he completed his Ph.D. at the Computer Laboratory of
the University  of Cambridge. His Ph.D.  research focused on ATM-based
site interconnection issues. This work, which was conducted within the
Unison  Project, led to  the early implementation of an ATM-based wide
area testbed.

Derek  McAuley  is  a  Lecturer  in  the  Computer  Laboratory at  the
University  of  Cambridge. His research  interests include networking,
distributed   systems   and   operating  systems.    Recent  work  has
concentrated on the support  of time dependent  mixed  media types  in
both  networks and operating systems. He has failed to leave Cambridge
since  arriving in  1979 to read Mathematics. In 1989 he completed his
Ph.D. on ATM internetworking. He has had a hand in  de-commissioning 4
ATM  networks,  including  Tennenhouse's carefully  constructed Unison
platform.
 
              
			   ---------------
			   L o c a t i o n
			   ---------------

The conference will be held in the Edward Lewis  Lecture Theatre which
is located in the  Windeyer Building on the UCL campus.  This building
is located on the corner of  Cleveland Street and Howland Street, with
the entrance  on  Cleveland Street.  Tutorials are all in UCL Computer
Science  Department in the Pearson Building, except T4 (Van  Jacobson)
on  the Tuesday which is held in the Edward Lewis Lecture Theatre. 

The main entrance of UCL  is located at the north end of Gower Street,
close to Euston Square, Warren Street,  or Euston tube stations.   The
UCL  Computer Science  Department is located in  the  basement of  the
Pearson Building.  Location


		     ---------------------------
		     T r a n s p o r t a t i o n
		     ---------------------------

* Getting to  London 

There  are  four  airports  in  and  around   London.   Here  is  some
information that  might help you to plan your journey.  Please consult
your travel agency or the airports directly for further information.

LONDON Heathrow Airport: 24 km west of London 
Telephone: +44-81-745-6156 
LONDON Gatwick Airport: 46 km south of London 
Telephone: +44-293-535-353 
STANsted Airport: 55 km north east of London 
Telephone: +44-279-680-500 

* Getting to UCL and Hotels

UCL is located  in central London, and is served by Warren St,  Euston
and Euston Square Underground (tube) stations, as well as several main
bus  routes.   The department  of  computer  science is  right  by the
entrance to the main quadrangles, on Gower Street.

From Heathrow:  Best  by  tube with Victoria Line  to  Euston  Station
(about #3, 50 minutes). Alternatives are via Bus with London Transport
A1 Airbus to Victoria Station (45 minutes).

For local hotels it is probably best to go to Euston Station and get a
taxi from  there  unless  you have a  street map already  and know the
nearest tube station.  A free tube map may  be obtained at any  ticket
office.

From Gatwick:  Best  by train, BR  Gatwick Express  to London Victoria
Station every 15 minutes (about #8.60, 30 minutes).

Unless you plan to sightsee  outside London a car  is probably a waste
of time.  Tube  fares are based on a zone system. After 9:30AM you can
get One Day Travel cards which allow you unlimited travel within given
zones  for the rest of the day -  that includes train and bus services
within  that zone too.  Zones 1,2 & 3 #2.30 pounds.   Zones  1-5 #2.60
pounds.


		      -------------------------
		      A c c o m o d a t i o n s
		      -------------------------

The following hotels are  walking distance from the conference meeting
room  on  the  UCL  campus.   Contact  the  hotel  directly  to  place
reservations.It  is highly recommended that reservations  are  made as
early as possible. Refer to SIGCOMM'94 when making the reservation.

* 	Hotel Ibis Euston
	3 Cardington Street, NW1
	Telephone: +44-71-388-7777, Fax: +44-71-388-0001
	Total Rooms: 300
	Single Room #49.50, Double Room #49.50
	Near UCL, about 10 minute walk from main Conference Hall. 

* 	St. George's Hotel 
	Langham Place, W1N
	Telephone: +44-71-580-0111, Fax: +44-71-436-7997
	Total Rooms: 86
	Single Room: #80.00, Double Room: #100.00 
	(Includes Continental Breakfast)
	Situated near Oxford Circus, about 10 minute walk from main venue. 

* 	RAMSAY HALL 
	20 Maple Street, W1P
	Total Rooms: 400
	Telephone: +44-71-387-4537, Fax: +44-71-383-0843
	Single Room: #19.50, Double Room: not available.
	(Includes Continental Breakfast) 
	Student residence used as hotel during summer break, 5 minute walk 
	from main conference venue.

* 	Hotel Russell 
	Russell Square, WC1
	Telephone: +44-71-837-6470, Fax: +44-71-837-2857
	Total Rooms: 328
	Single Room: #70.00, Double Room: #90.00 
	(Includes Continental Breakfast)
	Old Victorian Style Hotel. About 15 minute walk from Conference 
	venues. Russel Square  Station is on the Picadilly  line which
        reaches	to Heathrow Airport. Airport Bus stop nearby as well. 

* 	Forte Crest Bloomsbury 
	Coram Street, WC1
	Telephone: +44-71-837-1200
	Fax: +44-71-837-5374
	Total Rooms: 230
	Single Room: #69.00, Double Room: #79.00 
	(Includes Continental Breakfast)
	Modern hotel near Hotel Russell.

There  are  a large  number  of hotels near the conference. Almost any
hotel in the WC1 area of London is within 15 minutes walking distance.
A    list    of    more    hotels    may    be     found    via    www
(http://www.cs.ucl.ac.uk/sigcomm94)       or       anonymous       ftp
(norman.eng.buffalo.edu:/pub/SIGCOMM94). The list also includes nearby
lower cost housing and youth hostels.


	   ------------------------------------------------
	   R E G I S T R A T I O N    I N F O R M A T I O N
	   ------------------------------------------------

Full conference  registration  includes breaks, lunch, Tuesday evening
reception, one  ticket to dinner in the  Dinosaur Room of the  Natural
History Museum on Wednesday, and a copy of the conference proceedings.

Student  registration includes breaks, lunch and proceedings but  does
not include the  dinner/museum event.  On site registration will begin
Monday August 29, 1994 from 7:30AM - 5:00PM, and every day of the con-
ference starting at 7:30 am.



ACM and SIGcomm Membership
--------------------------

If  you are not an ACM or a SIGCOMM member at this time, you may  join
now to take  full advantage of ACM/SIGcomm Member or Student rates for
SIGCOMM94:

- ACM Associate Member Dues      	$82/#52
- Add SIGCOMM to ACM Membership      	$22/#15
- ACM Student Dues                  	$25/#17
- Add SIGCOMM to ACM Student Membership	$15/#10
- SIGCOMM Membership only (non-ACM)  	$50/#32

Total Membership Fees              $/#  _________

(Note: $ indicates U.S. dollars, and # British Pounds Sterling)

To advance the sciences and arts of information processing; to promote
the free interchange of  information about  the sciences  and  arts of
information processing both among  specialists and  among the  public;
and  to  develop  and  maintain  the  integrity  and   competence   of
individuals engaged  in  the  practice  of  information processing.  I
hereby affirm that I subscribe  to the purpose of  ACM and  understand
that my membership is not transferable.

Signature _________________________________________ Date ____________


Tutorials 
---------

Check each tutorial attending.  The tutorial registration fee includes
one copy  of the tutorial notes and  lunch.  Tutorials  are on a first
come first serve basis.

- T1 	Personal Communication Services & Networks (Monday) 
- T2 	Protocol Performance (Monday)
- T3 	Fiber Optic Networks (Tuesday) 
- T4 	Multimedia Conferencing on the Internet (Tuesday) 
- T5 	Asynchronous Transfer Mode (Tuesday) 
- W1 	Workshop on Distributed Systems (Tuesday)  

Tutorial Rates         
			Postmarked by         	Postmarked
			aug/1/1994           	after aug/1/1994
                                          
ACM/SIG Member          _____@ $275/#172        _____@ $325/#205  
Non-Member              _____@ $350/#220        _____@ $400/#250 
Student                 _____@ $138/#87         _____@ $175/#110 

Total Tutorial Fees     _____$/#                _____$/#  


Special Needs 
-------------

Vegetarian Meals:    	- Yes  	- No


Conference Registration
-----------------------

Please complete and  send  registration form, with check,  credit card
information or money orders (no purchase orders) to the address below.
Registrations accepted  via postal  mail,  fax or  email (with  credit
card) only.

				Postmarked by         	Postmarked
				Aug/1/1994           	after Aug/1/1994
                                          
ACM/SIG Member			_____@ $315/#200 	_____@ $365/#230  
Non-Member     			_____@ $397/#252  	_____@ $440/#275 
Student       	 		_____@ $100/#63 	_____@ $130/#82 
	
Total Registration  Fees    $/# _____               $/# _____

Extra Dinner/Museum Ticket      _____@ $55/#35


TOTAL ENCLOSED             $/#  _____ (ACM/SIGCOMM Membership, tutorials,
                                       conference registration)



NAME _________________________________________________________________

AFFILIATION __________________________________________________________

ADDRESS ______________________________________________________________

______________________________________________________________________

PHONE _____________________________ FAX ______________________________

Email ADDRESS ________________________________________________________

SIGCOMM Member? 	- YES    	 - NO
ACM/SIGCOMM Member Number ____________________________________________

CREDIT CARD PAYMENT 	- VISA   - MASTERCARD   - euroCARD

CARD HOLDER NAME _____________________________________________________

CARD NUMBER ______________________________________ EXP. DATE _________

SIGNATURE ____________________________________________________________


Please send this  form and a check,  credit card information  or money
orders (no purchase orders) to SIGCOMM'94.  Registrations accepted via
postal mail, fax or email only.

Send U.S. or   				Send Pound Sterling
Credit Card Payments to:

Patrick McCarren                        Soren-Aksel Sorensen
ACM - 17th Floor                        Dept. of Computer Science
1515 Broadway                           University College London
New York, NY 10036                      London WC1E 6BT
USA                                     United Kingdom
phone: +1 212/626/0611                  phone: +44 71 380 7269
fax: +1 212/302-5826                    fax +44 71 387 1397
mccarren@acm.org 

Email registrations  can  only  be  made  by a credit card  during the
pre-registration period ending 1 August 1994 and must use credit  card
payment.   A registration confirmation  letter  will  be sent to  each
participant  upon  receipt of  the  completed  registration  form  and
accompanying  payment.   Registration fee  will  be  refunded,  less a
$30/#19 administration fee, if  cancelation  notification  is received
prior  to  15  August  1994.  Substitution  for  a  paid  attendee  is
acceptable.
				   
	    ----------------------------------------------
	    C o n f e r e n c e    O r g a n i z a t i o n
	    ----------------------------------------------

General Chair:   Jon Crowcroft, University College London
Program Chairs:  Stephen Pink, Swedish Institute of Computer Science
                 Craig Partridge, BBN (Program Co-Chair for North America)

Ian F. Akyildiz, Georgia Institute of Technology, USA
Lillian N. Cassel, Villanova Univ., USA
Vinton Cerf, MCI, USA
Lyman Chapin, BBN, USA
Jon Crowcroft, Univ. College London, UK
Andre Danthine, Univ. of Liege, Belgium
Gary Delp, IBM, USA
Patrick W. Dowd, SUNY/Buffalo, USA
Deborah Estrin, Univ. Southern California, USA
David Feldmeier, Bellcore, USA
Sally Floyd, Lawrence Berkeley Laboratory, USA
David Greaves, ORL Cambridge, UK
Per Gunningberg, Swedish Institute of Computer Science, Sweden
Christian Huitema, INRIA, France
David Hutchison, Lancaster Univ., UK
Raj Jain, Ohio State University, USA
Jim Kurose, Univ. of Massachusetts, USA
Ian Leslie, Univ. of Cambridge, UK
David Oran, Digital Equipment Corp, USA
Gerard Parr, University of Ulster, Northern Ireland
Guru Parulkar, Washington Univ. St Louis, USA
Krzysztof Pawlikowski, Univ. of Canterbury, New Zealand
Bernhard Plattner, ETH Zurich, Switzerland
Scott Shenker, XEROX PARC, USA
Deepinder Sidhu, Univ. of Maryland-BC, USA
Jonathan M. Smith, Univ. Pennsylvania, USA
Khosrow Sohraby, Univ. of Missouri - Kansas City, USA
James Sterbenz, IBM Research, USA
Greg Watson, Hewlett Packard Labs, UK
Greg Wetzel, AT&T Bell Laboratories, USA
Lixia Zhang, XEROX PARC, USA


	 ---------------------------------------------------
	 F O R   A D D I T I O N A L   I N F O R M A T I O N
	 ---------------------------------------------------

Additional information may be found/requested from:

www:           http://www.cs.ucl.ac.uk/sigcomm94
anonymous ftp: norman.eng.buffalo.edu:/pub/sigcomm94
email:         sigcomm94@eng.buffalo.edu
fax:           +1 716.645.3656
phone:         +1 716.645.2406






-----------[000224][next][prev][last][first]----------------------------------------------------
Date:      Thu, 12 May 1994 13:11:00 GMT
From:      rstevens@noao.edu (W. Richard Stevens)
To:        comp.protocols.tcp-ip
Subject:   Re: ICMP timestamp, optional?

>    Support for the ICMP Timestamp messages is optional in the DoD
>    Internet protocol suite, so it is not surprising that most hosts
>    and gateways do not support it.
>
> Is it really optional?  Wouldn't this mean that users of ICMP
> timestamps, such as the BSD timed server, would be of limited use?
> Where can I find the answer?

Try it.  Pick up the program "icmptime" from ftp.uu.net in the file
published/books/stevens.tcpipiv1.tar.Z.  It sends an ICMP timestamp
request to a specified host, then calculates and prints the RTT.
Try a bunch of different hosts.  All systems with a TCP/IP stack
derived from the Berkeley sources should support it.  Some systems
(e.g., Solaris 2.x) make it configurable (and Solaris defaults
ip_respond_to_timestamp to 0, disabling the feature).

	Rich Stevens

-----------[000225][next][prev][last][first]----------------------------------------------------
Date:      12 May 1994 14:18:28 GMT
From:      mjr@tis.com (Marcus J Ranum)
To:        comp.protocols.tcp-ip,comp.protocols.tcp-ip.ibmpc
Subject:   Re: IP accounting program?

In article <2qlqfn$b52@ticsa.com> chris@ticsa.com (Chris Pinkham) writes:
>I'm looking for a program that will watch an Ethernet network and tell me
>how many bytes passed to/from particular networks to/from other networks
>in a measured period (e.g. one month/week/day etc.).

	You might want to look at NNStat -- available for FTP from
various places out there. It's pretty powerful and lets you have a
great deal of control over what kinds of information you want to
summarize.
	Actually doing charge-back from the summaries is another
matter -- I don't know of any systems for doing that.

mjr.

-----------[000226][next][prev][last][first]----------------------------------------------------
Date:      12 May 1994 16:59:46 GMT
From:      philb@cats.ucsc.edu (Philip Brown)
To:        comp.unix.solaris,comp.protocols.tcp-ip
Subject:   Re: SOCK_DGRAM bug in Solaris

Rob Stevens (robs@join.com) wrote:
: If you request a read of more bytes than are in the datagram
: you get only what was in the datagram (that's the meaning
: of a protocol which preserves record boundaries). If you
: request a read of fewer bytes, you get what you ask for and
: the rest is discarded.
 
: Regardless of whether you agree with the last sentence,
: that *is* the behavior in Solaris1 (specifically SunOS4.1.2).


Sounds like the "bug" is in sunos4 to me :-/

disclaimer: I am not a kewl soket haker.


-----------[000227][next][prev][last][first]----------------------------------------------------
Date:      12 May 1994 19:47:16 GMT
From:      scotta@kije.gsfc.nasa.gov (Scott Austin)
To:        comp.protocols.tcp-ip
Subject:   Re: Bargains on Comer's 3 Volume TCP/IP set

Corey Thompson <coreyt@vax1.mankato.msus.edu> wrote:

>  Would someone repost the original message, or tell me where I can get
>these 3 books?  Thank you.

Done.   Via email, that is.

Scott the-original-poster Austin
scott@cnt.com



-----------[000228][next][prev][last][first]----------------------------------------------------
Date:      Thu, 12 May 1994 21:19:3 GMT
From:      kiwilkra@inet.uni-c.dk (Lars Kraack)
To:        comp.protocols.tcp-ip
Subject:   Winsock spec - where?

I want to write a communications application that sits on top of a 
Winsock platform. Where can I get a Winsock spec and/or sample 
application programs using Winsock.

I am thinking of writing the application in Visual C or Borland C for 
Windows.

Regards, Lars

-----------------------------------------------------------------------
Lars Kraack
email      : kiwilkra@inet.uni-c.dk
CompuServe : 100273,3556
phone      : +45 38 887 772
-----------------------------------------------------------------------


-----------[000229][next][prev][last][first]----------------------------------------------------
Date:      Thu, 12 May 1994 21:43:43 GMT
From:      durham@mpr.ca (Paul Durham)
To:        comp.protocols.tcp-ip,comp.sys.m68k
Subject:   Looking for Ethernet-TCP/IP sw for Motorola 68360

I am looking for Ethernet-TCP/IP software for use on a 68360 or similar
M68K embedded systems. Any pointers would be much appreciated.


Thanks,

Paul
-- 
Paul Durham                    |      durham@mprgate.mpr.ca
MPR Teltech Ltd                | durham%mprgate.mpr.ca@ubc-cs.uucp
8999 Nelson Way                | ...uunet!ubc-cs!mprgate.mpr.ca!durham
Burnaby, B.C., Canada. V5A 4B5 |      phone: (604) 293-6035    

-----------[000230][next][prev][last][first]----------------------------------------------------
Date:      13 May 1994 08:41:10 -0600
From:      sbkumar@nyx10.cs.du.edu (suresh kumar)
To:        comp.programming,comp.client-server,comp.protocols.tcp-ip,comp.graphics
Subject:   Client  -  Server Development for Graphic Applications



Hi all:

I need to develop a client - server application. The server is going
to be a statistical or some database system which takes in some data
and plots some graphs and gives it back to the client. It will be
running on an Unix machine. The client will be either on a pc or some
Unix workstation. The easiest thing would be to transfer the data from
the server to the client nad plot it on the client's machine, but i dont
want the client (the end user) to buy some software to do it, as it will
add the cost.
Can anyone throw some light as how i can go about it -- the easiest
and fastest way to transfer the plotted file (a GIF file) from the 
server to the client. I am not sure with the limitations of using
sockets. Any advice will be highly appreciated.

Thanks
suresh


-----------[000231][next][prev][last][first]----------------------------------------------------
Date:      Thu, 12 May 1994 23:59:01 GMT
From:      itv@netcom.com (Harry G. Clayton)
To:        comp.protocols.tcp-ip
Subject:   Print Server Boxes

I'm trying to set up an Asante PrintServer box to connect a LocalTalk printer
to the TCP/IP network at my site.  I have no documentation.

Can anyone help?

I'm running SCO Unix here, and I'm no guru.  I figure I need to figure out
the IP address of the box somehow.  Is there anyway to "ping" it from the
server?  I also figure I need to set up a remote network printer.  I've got
the SCO manuals to do this, but can anybody verify that this is the right
way to go?

Thanks,

John J. Wilber
de facto Network Manager
ITV
itv@netcom.com



-- 
--------------------------
Harry G. Clayton		In CA:	Harry G. Clayton
ITV Technologies, Inc.			ITV Technologies, Inc.
17 Summer Hill Road			6800 Owensmouth Ave., Suite 230
Seymour, CT  06483-3535			Canoga Park, CA	 91303
USA					USA
Tel:	+1 203 888 5505			Tel:	+1 818 883 6333
Fax:	+1 203 881 9192			Fax:	+1 818 883 6692
E-mail: 72325.654@CompuServe.com   or	72234.2565@CompuServe.com

-----------[000232][next][prev][last][first]----------------------------------------------------
Date:      13 May 1994 08:26:49 -0500
From:      moswald@netserv.unmc.edu (Mike Oswald)
To:        comp.protocols.tcp-ip
Subject:   TCP/IP, SLIP Book Recommendations Needed

Anyone have any TCP/IP & SLIP book recommendations for a newbie?

Thanks.

-Mike

-----------[000233][next][prev][last][first]----------------------------------------------------
Date:      Fri, 13 May 94 11:54:03 -0700 (PDT)
From:      KCARPENT@mindlink.bc.ca (Ken Carpenter)
To:        comp.protocols.tcp-ip
Subject:   Multi-homed hosts & Subnet Masks

I am working on an embedded application using UDP/IP for communications and
have come up against a bit of a problem.

Our embedded systems can have from 1-5 different network connections, and
under the IP addressing scheme, each adaptor has its own IP address, not
each host.

To simplify routing, I want to be able to address the host with the same
address regardless of which adaptor the packet will eventually end up
arriving on.

It has been suggested to me that this should be possible using an
appropriate Subnet mask.

Can anyone enlighten me as to the correct method of accomplishing this?

Thanks,

Ken Carpenter
Research & Development
Delta Controls

-----------[000234][next][prev][last][first]----------------------------------------------------
Date:      Fri, 13 May 1994 13:20:12 -0500
From:      ricketts.je@shocker.ee.twsu.edu (Ricketts, Joel E.; EE-U)
To:        comp.protocols.tcp-ip
Subject:   Where can I get Internet's RFC's?

Where would I go about getting Internet's Request for Comment files/
documents (RFC's)?  I have been looking for technical information relating
to these files, but have had no luck.
Joel

-----------[000235][next][prev][last][first]----------------------------------------------------
Date:      13 May 94 07:35:36 GMT
From:      oscom@yarrow.wt.uwa.edu.au (OSCOM International)
To:        comp.protocols.tcp-ip,comp.protocols.tcp-ip.ibmpc
Subject:   WATTCP : FTP problem

Hi 

I am using the WATTCP FTP program version 0.6 and I am experiencing
weird problems with the program...!

I can download data to the HP workstation at 300Kbytes/sec BUT i upload
at only 1Kbyte/Sec which is poor quality as far as anyone is concerned!!

Has anyone out there had similar problems with WATTCP FTP program... and
if you have managed to fix the problem please drop in a note ..


Thanx in advance..
Chris Avis

oscom@yarrow.wt.uwa.edu.au




-----------[000236][next][prev][last][first]----------------------------------------------------
Date:      Fri, 13 May 1994 08:00:32 GMT
From:      johnr@kuentos.guam.net (John Rhee)
To:        comp.protocols.tcp-ip,comp.protocols.tcp-ip.ibmpc,comp.protocols.ppp
Subject:   [Q]PPP Packet Driver for WinSock


I am looking for the PPP packet driver it works with Windows Socket packages.

Now, I am using SLIPPER. I would like to move to PPP.

Bye...

----------------------------------------------------------------------------
John Rhee
Johnr@Kuentos.Guam.NET				        671-632-1225 : Voice
Kuentos Communications, Inc.			        671-632-5641 : Fax
PO Box 22887
GMF, GUAM 96921
----------------------------------------------------------------------------

-- 
----------------------------------------------------------------------------
John Rhee
Johnr@Kuentos.Guam.NET				        671-632-1225 : Voice
Kuentos Communications, Inc.			        671-632-5641 : Fax

-----------[000237][next][prev][last][first]----------------------------------------------------
Date:      13 May 1994 08:23:07 GMT
From:      bortz@cnam.cnam.fr (Stephane Bortzmeyer)
To:        comp.protocols.tcp-ip
Subject:   SUMMARY: FTP problem between Europe and USA

In article <2qqqdc$qec@sheckley.cnam.fr>, bortz@cnam.cnam.fr (Stephane Bortzmeyer) writes:
>We have a strange problem with ftp transfers.
>The problem appeared two weeks ago. Before that, *all* ftp transfers 
>went fine, either with France or with the USA, whatever the size of 
>the files were. But now, we are unable to transfer some files from 
>some US anonymous ftp servers.
>The symptom is: not one byte comes to us, and we have a TCP reset a 
>long time after.

Well, the problem is apparently solved. At least, it worked this morning 
on all the sites I mentioned.

Our network provider told us there was a problem on the transatlantic 
line and they fixed it. As usual, they didn't give any details (the 
provider is France-Telecom, which operates the Renater network).

Now, the question is: can anynone explain why a f*$#@ problem on a 
line, instead of resulting in loss packets, delays, etc, results in 
impossibility of transferring *some* files and not others? And why
not even one byte is transferred on the offending files?

Thanks a lot to Paul Svensson <paul@lysator.liu.se>, Kurt Jaeger 
<Kurt.Jaeger@rus.uni-stuttgart.de > and Jean-Michel Planche 
<jmp@apysoft.oleane.com> for testing their connections and proving 
the problem was specific to the Renater network.

Stephane Bortzmeyer           Conservatoire National des Arts et Metiers	
bortzmeyer@cnam.fr            Laboratoire d'Informatique
                              292, rue Saint-Martin			
tel: +33 (1) 40 27 27 31      75141 Paris Cedex 03
fax: +33 (1) 40 27 27 72      France	

"C'est la nuit qu'il est beau de croire a la lumiere." E. Rostand

http://web.cnam.fr/personnes/bortzmeyer/home_page.dom

					
	

-----------[000238][next][prev][last][first]----------------------------------------------------
Date:      Fri, 13 May 94 10:39:16
From:      John.Mccormack@hrp.no
To:        comp.protocols.tcp-ip
Subject:   Re: FTP problem between Europe and USA



> We have a strange problem with ftp transfers......
 
> The symptom is: not one byte comes to us, and we have a TCP reset a 
> long time after.
> 
> Examples:
> 
> ftp://ftp.netcom.com/pub/livingston/sun/pm2_release_3.0.3.tar (the 
> file is 1,4 Mbytes, the server is wu-ftpd "2.0WU(10)")
> 
	I received this one no problem.  I didn't try the others.

Halden,  Norway.


-----------[000239][next][prev][last][first]----------------------------------------------------
Date:      13 May 1994 11:53:51 GMT
From:      sydtran@wam.umd.edu (Sydney Quocsi Tran)
To:        comp.protocols.tcp-ip
Subject:   OSPF RFC?

Can anyone please tell me what is the RFC number for OSPF routing protocol
and where I can find it?  I was informed that RFC1131 (OSPF) is only
available in post script format and it's not the latest RFC either.
Your help is greatly appreciated.  Thanks
.


-----------[000240][next][prev][last][first]----------------------------------------------------
Date:      Fri, 13 May 1994 12:59:33 GMT
From:      clairmon@spectrum3_le0.ml.com (J. Clairmont)
To:        comp.protocols.tcp-ip,comp.protocols.snmp
Subject:   SNMP Management Platforms

                              
Following is a feature list comparing some of the SNMP management
platforms available.  These are the ones that we evaluated.
I was wondering if anyone else has done this and if they wish to comment?  I
would appreciate any comments and/or disagreements with the list.

The current feature list is of the software that was
available as of 4/1994.  Any feature additions since then have not been noted.
This is a preliminary list and is not a complete list of the features of the
following products.  This is essentially a list of my opinions as to
what these products offer.  Dissenting views are encouraged for a fair 
evaluation.

Trakker - SNMP management platform running in conjunction with SunNet Manager.
Has extensive reporting capabilities,  has an SQL Database(INGRESS),
is slated to be relational database independent.  Has a good historical and
realtime reporting mechnanism.  Has a SAS gateway for SAS/CPE.  The reporting 
capabilities are quite extensive and with SQL it gives a good AD HOC and textual
report capability.  The Concord Probes look at all packets, all dialogs seen are
kept.  It keeps a packet count, byte count by protocol(decodes oever 70),
segment pairs, MAC, name(if available), router type.  Fairly easy to use, all
the reports are scriptable in sh, and available for cron execution.  Has all the
information needed to do Billing/Tariffing on a LAN backbone.

 Problems: No RMON though they will have it soon, need to do Banyan decoding.  
 May give too much information, right now it has a proprietary MIB.   Does not
 do any Reporting of other device's MIBS,  no WAN/MAN performance reporting.
 
HP/NetMetrics/EASE/pmReporter - Does protocol decoding, reporting mechanism are
extensive, though at this time not integrated.  Currently, a collection of three
products.  The LAN probe is remotely configurable and downloadable, for a large
network this could save a lot of headaches.  All the data is sampled, every 
400 packets or so.  It would not be easy to do billing/tariffing because it uses  
sampled data, the inability to do logical groupings by departments
and some other issues( such as keeping track of Dialogs, etc).  This is soon
to be rectified and am looking forward to the final integrated product.
Most of the reports are not scriptable, so a screen capture by xwd or SNAP
must be done.   Very nice graphical interface, but because it is three 
separate products, it is not easy to use.

Cabletron Spectrum - Huge graphical interface, very nice looking.  But is
difficult to manage and use.   Requires lots of support, but has an excellent 
SAS/CPE gateway. Good Alarming capabilities, difficult to learn and use.  
No protocol decode, some reporting capability(BETA ver. more), graphical reports are 
basically RMON.  This is not an easy system to evaluate, but is overall too
complex for easy evaluation or use.  It requires lots of resources with a large 
network 20000+ stations.  It needs a Sparc 10 with 128 meg to run and 1 gig. of
disk space.  But it is interesting to use, though definitely not for novices.

XSniffer - every tech. loves the SNIFFER, this is the distributed Sniffer.  Does
all the great things Sniffer does but on a Sun.  But its Network Reporting,
historical trending and other things do not make it a complete SNMP Network
Performance Management tool.  Good things are yet to come, though, I am sure.

NetScout - Nice little box, inexpensive, good for remote segments.  No real
enterprise wide reporting.  No historical database, reporting somewhat limited.
Not quite adequate for Large Network Management Issues.
  
We also looked at WG, and some others but, settled on these to evaluate because
of the Users particular needs.  Again any comments are appreciated!

  Report as of 5/12/1994 - SNMP Network Management  Feature
                 Survey by Jan M. Clairmont
------------------------------------------------------------------
| Feature | Trakker |  HP | XSNIFFER | Cabletron | NetScout      |
------------------------------------------------------------------
Multiple    Yes       Yes     Limited     Yes        No      
users ?                        
------------------------------------------------------------------
Use         Yes       Yes       Yes       Yes        No
multiple
tools
concur-
renntly?
------------------------------------------------------------------
Real        Yes       Yes       Yes        No     Limited
Time
Monitoring of
specific
protocols
layers?
------------------------------------------------------------------
Real        Yes       Yes       Yes        No     Limited
time
trouble
shooting
of
higher
level
protocols?
------------------------------------------------------------------
AutoMapp    Yes       Yes        No       Yes        No
ing of
Network
Topology?
------------------------------------------------------------------
Partial     Yes       Yes        No       Yes        No
Auto
Remapping?
------------------------------------------------------------------
Historical  Yes       Yes     Limited   SAS/CPE      No
                                 to
Analysis                      Sniffer
?                               Mem.
------------------------------------------------------------------
View        Yes       Yes       Yes       Yes       Yes
Multiple
Segments?
------------------------------------------------------------------
Network     Yes       Yes       Yes       Yes       Yes
Wide
Thresholds?
------------------------------------------------------------------
Graphical   Yes       Yes       Yes     Limited     Yes
Reports?
------------------------------------------------------------------
Scriptable  Yes        No        No        No        No
Automatted
Reporting?
------------------------------------------------------------------
DOS         Yes       Yes       Yes        No       Yes?
/Windows
Support?
------------------------------------------------------------------
Remote      Yes       Yes        No        No        No
Probe
Download-
Firmware?
------------------------------------------------------------------
Remote      No        Yes        No        No        No
Probe
Download
Configuration?
------------------------------------------------------------------
Graphical   Yes       Yes       Yes     Limited     Yes
Reports?
------------------------------------------------------------------
Textual     Yes     Limited   Limited   Limited   Limited
Based
Reporting?
------------------------------------------------------------------
Relational  Yes        No        No     DB Vista     No
Database
Support?
------------------------------------------------------------------
Probe       Yes       Yes       Yes        No       Yes
Serial
Support?
------------------------------------------------------------------
Protocol                                              
Decode
Supported:  Yes       Yes       Yes        No       Yes
------------------------------------------------------------------
TCP/IP?     Yes       Yes       Yes        No       Yes
------------------------------------------------------------------
NFS?        Yes       Yes       Yes        No       Yes
------------------------------------------------------------------
DECNet?
            Yes       Yes       Yes        No       Yes
------------------------------------------------------------------
LAT?
            Yes       Yes       Yes        No       Yes
------------------------------------------------------------------
IPX?
          Limited   Limited   Limited      No     Limited
------------------------------------------------------------------
Banyan
Vines?
            Limited   Yes       Yes        No       Yes
------------------------------------------------------------------
Appletalk?
            Yes       Yes       Yes        No       Yes
------------------------------------------------------------------
SNA?        Yes       Yes        No       Yes        No
------------------------------------------------------------------
SNMP        Yes       Yes        No       Yes        No
traps and
alarms?
------------------------------------------------------------------
Reporting:

------------------------------------------------------------------
Logical
Grouping?   Yes       Limited    No        No        No
------------------------------------------------------------------
  
Trend
Analysis?
            Yes       Yes        No        No         No   
------------------------------------------------------------------
Protocol
Distribu
tion?
            Yes       Yes     Limited      No        Limited
------------------------------------------------------------------
Bandwidth
Utitliza
tion?
            Yes       Yes     Limited      No        Limited
                                        SAS/CPE
------------------------------------------------------------------
Exception
Reporting?
          Limited   Limited      No         No         No
------------------------------------------------------------------
Modeling -
What If
Reporting?  Prototype  Limited   ?           No        No
------------------------------------------------------------------
Supports
SAS
CPE?        Yes        No        No         Yes         No
------------------------------------------------------------------
Database
Rollups?    Yes        No        No          No        No
------------------------------------------------------------------
Tariffing
and/or
Billing
Support?    Yes        No        No          No        No
------------------------------------------------------------------
Decode     
PreFilters? Yes       Yes        Yes         No        Yes
------------------------------------------------------------------
Decode      Yes       Yes        Yes         No        Yes
PostFilters?
------------------------------------------------------------------
Expert    
Help on
Network
Problems?  Limited   Limited   Limited       No         No
------------------------------------------------------------------
Ease of     Yes       Not       Yes          No        Yes
Use?
------------------------------------------------------------------
PostScript  Yes        No        No          No         No
Printing
Support?
------------------------------------------------------------------
Help       Almost      No        No          No         Yes
Menu
that is
Helpful?
------------------------------------------------------------------
RMON      Not Yet     Yes        ?         SAS/CPE      ?
Support?  Projected                        Limited
------------------------------------------------------------------
MIB         Yes       Yes       Yes           No        No
Extensions?
------------------------------------------------------------------
Software    Yes        ?         No           No        ?
Probe
Available?
------------------------------------------------------------------
100mega-    No       Yes        No            No         No
bit               NetMetrics
FDDI                 Only
Support?          on Sun Sparc
-------------------------------------------------------------------
Create      Yes     Limited      No         Limited      No
Own Ad
Hoc
Reports?
-------------------------------------------------------------------
Programa-    Yes       No        No          Limited     No
ble API
access
to
Database
and
System?
-------------------------------------------------------------------
                                                      

---
--         /\   __
   |      // \_/_ \  Jan Clairmont  Miami Jan(Road Warrior)
   |  __ `'_[(.)*]\\ email:   jmc@devjam1.ml.com or clairmj@mary.iia.org
 \_|_(_(__/ )` @ ' `'(212) 236-2605
             (_V_) 
                                                       
                                                      
                                                      
                                                      
                                                      
                                                      
                                                      
                                                      
                                                      
                                                      
                                                      
                                                      
                                                      




-----------[000241][next][prev][last][first]----------------------------------------------------
Date:      13 May 1994 14:12:51 GMT
From:      cianci@novell.dima.unige.it (Roberto Cianci)
To:        comp.protocols.tcp-ip
Subject:   please suggestions for telnetd.exe  -  ioo [1/1]

Hi all,
i wish to receive suggestions for the installation of 
telnetd.exe daemon on a 386 machine. In our dept. we have
a netware 3.11 server and we wish to open some databases which
are on a directory of this server from a remote computer; 
unfortunately we have to use a clipper compiled program, so the unique 
possibility to use it 
from our vax or sun computers is   to telnet to a 386 where we have
installed ipx and telnetd:

my config.sys is very simple

device=qemm386.sys auto ram
files=70
dos=high


my autoexec.bat

path= ....
loadhi lsl
loadhi 3c523
loadhi odipkt    (release 3.0)
loadhi ipxodi
loadhi netx
login user
castoff
telnetd

my wattcp.cfg

 my_ip=130.251.160.195
 netmask=255.255.255.0
 telnetd.password=pass
 telnetd.message="Welcome"
 inactive=300

my net.cfg is

Link Support
        Buffers 8 1500
        MemPool 4096

Link Driver 3C523
        Frame Ethernet_II
        Frame Ethernet_802.3
        Protocol IPX 0 Ethernet_802.3
        Protocol IP 1 Ethernet_802.3

Protocol TCPIP
        ip_address      130.251.160.195
        ip_router       130.251.160.254
        ip_netmask      255.255.255.0
        tcp_sockets     8
        udp_sockets     8
        raw_sockets     1
        nb_sessions     4
        nb_commands     8
        nb_adapter      0
        nb_domain       


Now, i would like to receive suggestions
on these topics:

1 the keyboard is not good: where is the delete key ? for example
   how to remap keys?
2. the inactive switch does not work. why?
3. how to set up parameters in order to increase network performance? 

Has anybody which has already installed such a program examples and
suggestion for me?
thank you all
roberto cianci

Math. Dept. Univ. GENOA

-----------[000242][next][prev][last][first]----------------------------------------------------
Date:      Thu, 12 May 1994 19:09:56 +1200
From:      chris-b@cs.aukuni.ac.nz (Chris Burns)
To:        comp.protocols.tcp-ip
Subject:   MacTCP Development

Hi dudes,

We're just starting work on a TCP/IP based hypermedia project similar in
scope to "NCSA Mosaic". The Comp Sci Dept that I work in does not have much
up-to-date documentation about MacTCP from a programmer's viewpoint, so I
was wondering if you could help me out by pointing me in the right
direction.

I've seen stuff about "MacTCP 3.0" and the "Open Transport TCP Library"
(shared library manager). What gives? Is one of these the preferred way to
go, or do we stick with MacTCP 2.0.4?

I've found plenty of header files:

AddressXlation.h
DNR.c
GetMyIPAddr.h
MacTCPCommonTypes.h
MiscIPPB.h
TCPPB.h
UDPPB.h

But no libraries implementing the features found therein.
Help!!?!!

Any information will be treated in the utmost confidence and viewed with
great appreciation.

Thanks in advance for your time,

Chris B
-----------------------------------------------------------------------
NewZealand:AucklandUniversity:ComputerScience:HyperMediaUnit:ChrisBurns
Internet: chris-b@cs.aukuni.ac.nz
Phone:    +64 9 373-7599 x6137
Fax:      +64 9 373-7453
-----------------------------------------------------------------------

-----------[000243][next][prev][last][first]----------------------------------------------------
Date:      Fri, 13 May 1994 15:19:24 GMT
From:      jcd@aber.ac.uk (John Cormac Davis)
To:        comp.os.ms-windows.networking.tcp-ip,comp.protocols.nfs,comp.protocols.tcp-ip
Subject:   NDIS / Packet driver Shim

Hello,

I'm trying to get the MS TCP/IP beta working over a packet-driver interface.
Does anyone know of a NDIS over packet driver shim that might help,
( as opposed to the dis-pkt packet-driver over NDIS that I have already
come across) ?

Thank you ,

JD

-----------[000244][next][prev][last][first]----------------------------------------------------
Date:      13 May 1994 19:05:18 GMT
From:      bradley@wg.com (Doug Bradley)
To:        comp.protocols.tcp-ip
Subject:   how to RARP?


could any kind soul out there please point me to some reference
(or better yet example source) on how to make a rarp query?
rfc903 was not much help.

many tia,
d
--
--
Doug Bradley
Wandel & Golterman Technologies
Research Triangle Park, NC   27709-3585 	USA
919-941-5730					bradley@wg.com

-----------[000245][next][prev][last][first]----------------------------------------------------
Date:      Fri, 13 May 1994 20:43:36 GMT
From:      tbel@oahu.cern.ch (Tim Bell)
To:        comp.protocols.tcp-ip
Subject:   Forwarding a socket connection to another machine

Is there a way to forward a socket connection to another machine ? I
was to write a gateway program that will get called when a connection
is made but then forward on the connection to another machine (it
won't be the same machine each time).  Since a large amount of data
will be transferred, I don't want to put a load on the intermediate
machine so a simple read/write forwarder won't work... in pictures...


     A  ------->   B
			---------> C

     A --------------------------> C

Tim.


--
Tim Bell
IBM High Energy Physics European Support
E-mail: tbel@oahu.cern.ch Office: 513-R003 Phone: x7081

-----------[000246][next][prev][last][first]----------------------------------------------------
Date:      Fri, 13 May 1994 20:59:07 GMT
From:      thinman@netcom.com (Technically Sweet)
To:        comp.protocols.tcp-ip
Subject:   BOOTP: where for client?

Where would I find sample client code for the BOOTP protocol?
And, where would I find discussion of what it takes to implement
this under the BSD flavor of TCP/IP?  (This is for a port of
BSD TCP/IP into a monolithic real-time kernel, so kernel/user
space causality & design issues don't matter.)

Please just send mail and I will summarize.

Thanks,

-- 

Lance Norskog
thinman@netcom.com
Artisputtingtogether. Art  s th ow n  aw y.

-----------[000247][next][prev][last][first]----------------------------------------------------
Date:      Fri, 13 May 1994 21:01:30 GMT
From:      thinman@netcom.com (Technically Sweet)
To:        comp.protocols.tcp-ip
Subject:   Multiple gateway protocols: how?

A few years ago the preferred way to run multiple routine protocols
was the GATED program.  What routing protocols are prevalent these
days, and what publically available programs implement them?

Please just mail me and I'll summarize the responses.

Thanks,

-- 

Lance Norskog
thinman@netcom.com
Artisputtingtogether. Art  s th ow n  aw y.

-----------[000248][next][prev][last][first]----------------------------------------------------
Date:      13 May 1994 21:10:22 GMT
From:      jliv@wellfleet.com (John Loiacono IV)
To:        comp.protocols.tcp-ip,comp.dcom.sys.wellfleet,comp.dcom.sys.cisco
Subject:   Re: Variable IP subnets with RIP on Wellfleet

Yes, we do support variable length subnet masks.  We also have a large
number of customers who have implemented this feature with excellent
results.

As for CIDR, we plan to support this in a future release.  I will have
to defer to Prod. Mgmt., for specifics.

Regards,

John Loiacono IV
Wellfleet Communications, Inc.
jliv@wellfleet.com

-----------[000249][next][prev][last][first]----------------------------------------------------
Date:      13 May 1994 21:53:20 GMT
From:      zhijun@slinky.cs.nyu.edu (Zhijun Liu)
To:        comp.multimedia,comp.protocols.tcp-ip
Subject:   Is there any Sockets Library for Macintosh?

  I'm looking for some Sockets library building upon MacTCP. Is there 
  any such package available? Any relevant information is aprreciated.

  --Z. Liu
  zhijun@cs.nyu.edu

 



-----------[000250][next][prev][last][first]----------------------------------------------------
Date:      Fri, 13 May 1994 23:01:25 GMT
From:      albaugh@agames.agames.com (Mike Albaugh)
To:        comp.protocols.tcp-ip
Subject:   Re: Duplicate IP address detection

Dale R. Worley (drw@kutta.mit.edu) wrote:
: In article <2qmjr2$49b@charm.magnus.acs.ohio-state.edu>
 maf@dunedin.acs.ohio-state.edu (Mark Fullmer) writes:
:    In article <2qlq83$uq@raven.dow.com> Private_User@dow.com writes:
:    >
:    >Does anyone have a simple way of detecting duplicate IP address usage on
 a LAN
:    >!before! the offending node has done damage (like jamming NFS)?
 
: But basically, there's no way (within IP) to tell if a particular
: address is in use, except by attempting to send to it.  However, that
: requires that the sender have an IP address, to which the reply can be
: returned.

	He did say "On a LAN" which would possibly lead to strategies
using ARP. I know it's not a general solution, but it can help. Now if we
can only get rid of the printer-boxes that respond to random ARPS :-)

					Mike

| Mike Albaugh (albaugh@agames.com)
| Atari Games Corp (Arcade Games, no relation to the makers of the ST)
| 675 Sycamore Dr. Milpitas, CA 95035		voice: (408)434-1709
| The opinions expressed are my own (Boy, are they ever)

-----------[000251][next][prev][last][first]----------------------------------------------------
Date:      14 May 1994 19:39:45 -0700
From:      phil@lykos.netpart.com (Phil Trubey)
To:        comp.protocols.tcp-ip
Subject:   whois++ client implementations?

Does anyone know if there is working being done on non-unix whois++
clients?  Perhaps even work integrating whois++ into popular e-mail
clients like Eudora, or other commercial e-mail systems?

Thanks,

-- 
 Phil Trubey                 | Providing independent consulting in the 
 NetPartners                 |   application of Internet technology.
 E-mail: phil@netpart.com    |   Home Page: http://www.netpart.com/

-----------[000252][next][prev][last][first]----------------------------------------------------
Date:      14 May 1994 08:40:09 GMT
From:      bortz@cnam.cnam.fr (Stephane Bortzmeyer)
To:        comp.protocols.tcp-ip
Subject:   Re: SUMMARY: FTP problem between Europe and USA

In article <2qvddb$jbe@sheckley.cnam.fr>, bortz@cnam.cnam.fr (Stephane Bortzmeyer) writes:
...
>Well, the problem is apparently solved. At least, it worked this morning 
>on all the sites I mentioned.
 ...
>Now, the question is: can anynone explain why a f*$#@ problem on a 
>line, instead of resulting in loss packets, delays, etc, results in 
>impossibility of transferring *some* files and not others? And why
>not even one byte is transferred on the offending files?

It's specially strange since during the crisis, ping and traceroute 
both reported normal delays and losses. All the other cable problems 
I saw (either on Ethernet or on leased lines) were easily reported by 
these tools: high percentage of lost packets, incredible delays, etc.

So, nobody gets an idea? (I don't :-( )

Stephane Bortzmeyer           Conservatoire National des Arts et Metiers	
bortzmeyer@cnam.fr            Laboratoire d'Informatique
                              292, rue Saint-Martin			
tel: +33 (1) 40 27 27 31      75141 Paris Cedex 03
fax: +33 (1) 40 27 27 72      France	

"C'est la nuit qu'il est beau de croire a la lumiere." E. Rostand

http://web.cnam.fr/personnes/bortzmeyer/home_page.dom
					
	

-----------[000253][next][prev][last][first]----------------------------------------------------
Date:      Sat, 14 May 1994 09:27:30 GMT
From:      L.G.Cuthbert@qmw.ac.uk (Laurie Cuthbert)
To:        comp.protocols.tcp-ip
Subject:   a SLIP server

At present we have at QMW a central modem pool with a terminal server that 
will either allow a standard ASCII serial login or SLIP. The other side is 
connected to the Campus LAN so that we can then connect anywhere on (or off) 
site. The problem is that the SLIP implementation is really bad becuase of the 
way it implements rate adaptation in the direction from the ethernet to the 
SLIP port.

The obvious solution is to replace it and this will be done in the medium term.

In the short term we are desperately looking for alternative solutions. 

One theoretical possibility that would be worth a try (if possible) is to do 
an ASCII login to the remote workstation and then start SLIP on the 
workstation so that the terminal server is not doing the ethernet/SLIP 
conversion. The point is: is there any software that would do this on the 
remote workstation - it would need to run SLIP over the ETHERNET port (not the 
serial port) of the workstation.

Laurie Cuthbert

-----------[000254][next][prev][last][first]----------------------------------------------------
Date:      14 May 94 13:00:01 GMT
From:      chrisc@fir.canberra.edu.au (Chris Chlap)
To:        comp.protocols.tcp-ip
Subject:   Re: TCP/IP D.E. Comer Volume III

In <johnston.40.00061C34@hookup.net> johnston@hookup.net (Stewart Johnston) writes:

>TCP/IP   D.E. Comer Volume III
 
>I have just purchased "Internetworking with TCP/IP volume III" by Douglas E. Comer and David
>L. Stevens, published by Pentice Hall.  On the back of the book there is a note that reads:
>	"  - Includes programs available from the publisher and by FTP. "
 
>I assume this means the example code in this book is available through FTP.  Unfortunately the
>authors failed to publish the FTP sight address.
 
>Could someone please help me find the FTP address for the above source code.
 
>Your help would be greatly appreciated.
 
>Thanks
>Stewart Johnston
>johnston@hookup.net
Vol. 3 ia available in two versions(BSD and TLI). The BSD version is
available from ftp.uu.net in published/books (I can't remember the full
path name) and the TLI version is available from ftp.cs.purdue.edu in
/pub/dls.

Chris Chlap.

-----------[000255][next][prev][last][first]----------------------------------------------------
Date:      Sat, 14 May 94 14:48:50 GMT
From:      epoole@leotech.mv.com (Eric Poole)
To:        comp.protocols.tcp-ip,comp.protocols.tcp-ip.ibmpc
Subject:   TERMINAL SERVER Software for Linux - Where?

Can anyone direct me to some software that will allow a Linux box to
make like a terminal server?

Specifically, what I need is hardware/software so that Internoids can
telnet into here and access our Wildcat BBS ... they would telnet in
over TCP/IP (well, SLIP), then go out one of the Linux box's serial
ports to a user port on the BBS.

I plan to use a 386/33 or 386/40 for this, with 8 serial connections
to the BBS.

I know of at least one site (bbs.meu.edu) that is doing this using a
"real" terminal server, but I can't afford the $2000+ pricetag for one
of those little boxes.

Would greatly appreciate any help anyone can offer ... please respond
either here or by mail to epoole@leotech.mv.com, and I'll summarize
any e-mail replies.

When you reply, please keep in mind that I am still quite a Unix
novice ... thanks for your consideration <smile> ...

Thanks to all ...

 . . . . . ep



-----------[000256][next][prev][last][first]----------------------------------------------------
Date:      14 May 94 22:23:41
From:      billw@glare.cisco.com (William )
To:        comp.protocols.tcp-ip
Subject:   Re: a SLIP server

Many terminal servers support SLIP these days.

    The problem is that the SLIP implementation is really bad
    becuase of the way it implements rate adaptation in the direction
    from the ethernet to the SLIP port.

What does "rate adaption" have to do with slip and ethernet?

BillW

-----------[000257][next][prev][last][first]----------------------------------------------------
Date:      14 May 1994 20:10:01 GMT
From:      kruckenb@sal.cs.utah.edu (Joseph Kruckenberg)
To:        comp.os.os2.networking,comp.os.os2.networking.tcp-ip,comp.os.os2.setup,alt.internet.access.wanted,comp.protocols.tcp-ip,comp.protocols.tcp-ip.ibmpc,comp.protocols.ppp
Subject:   [Q]Using Linux or OS/2 as Internet Gateway/Firewall

[Followups to this post will go to comp.os.os2.networking.tcp-ip and
comp.os.linux.misc]

I'm checking into how the company I work for can get a live connection
to the Internet, and need some guidance.  We're running on a Novell
3.11/4.x network with Mac, DOS/Windows, and OS/2 clients.  We want to
provide Internet access for our users for email (probably gatewayed
through cc:Mail), Mosaic, gopher, telnet, ftp, Usenet news, and other
services.  We are probably going to use a dedicated dial-up service
(28.8K modems) via SLIP or PPP for now, with plans to move to
higher-bandwidth connections as demand increases.

Our plan now is to have both TCP/IP and IPX running on our network,
with a computer acting as a local Internet node on the network.  This
computer would be connected via dial-up or leased line to our Internet
provider.

I'm trying to figure out how we can use this computer to be both a
gateway (with some security) and a local repository for Usenet
newsgroups, possible external ftp/telnet access, etc.  We would be
running on our own domain (xyz.com), so I guess we'd allocate IP
addresses within our organization.  We want to be able to restrict
access to our network, such as only allowing in traffic for certain
addresses, or only at certain hours of the day or certain days of the
week.  This traffic wouldn't have the address of the gateway, but of a
client machine on the network.

However, there would be traffic for the gateway machine as well
(ftp/telnet/finger/UUCP-transer/other traffic).  Originating from our
network we would have outgoing traffic intended for other machines on
the Internet that would have to pass through the gateway, but we might
want to be able to limit that traffic as well to conserve the
bandwidth we have.

What kind of software would we have to get to be able to do this?  Is
it available for OS/2 or Linux?  Which platform is going to be easier
to set-up and more capable of doing these various tasks?  Can you
recommend some books or on-line resources (ftp, telnet, newsgroups,
etc) that I can use to learn how these types of things are done (I've
got Linux running via SLIP, so I'm not completely helpless about some
of these things).

I appreciate your time and help in getting this set up.  Hopefully my
company will let me use the connection to provide some services to the
Internet when we do get it running.

Thanks.
Pete Kruckenberg
kruckenb@sal.cs.utah.edu


-----------[000258][next][prev][last][first]----------------------------------------------------
Date:      Sat, 14 May 94 21:20:18 GMT
From:      mliu@garden.csc.calpoly.edu (Mei-Ling L. Liu)
To:        comp.dcom.lans,comp.protocols.tcp-ip,comp.protocols.misc
Subject:   WANTED: Vnet (VNET?) Info

I am looking for any information which can get me started with VNET,
a free software which I am told can be used for network class projects.
If your know where it can be ftp-ed from and/or have startup info,
I would greatly appreciate it.

Mei-Ling Liu
Distributed Systems Lab
UCSB


-----------[000259][next][prev][last][first]----------------------------------------------------
Date:      15 May 1994 09:37:23 GMT
From:      tchen@umich.edu (Antony Chen)
To:        comp.protocols.tcp-ip,comp.protocols.tcp-ip.ibmpc,comp.protocols.ppp
Subject:   Re: [Q]PPP Packet Driver for WinSock

John Rhee (johnr@kuentos.guam.net) wrote:

: I am looking for the PPP packet driver it works with Windows Socket packages.
 
: Now, I am using SLIPPER. I would like to move to PPP.

Ftp to merit.edu in /pub/pc/msdos/ppp or something like that.  Grab
the file etherppp.zip.  

--
***************************************************************************
MMMMM      MMMMM  Antony Chen @ University of Michigan, Ann Arbor         |
 MMMM      MMMM   ---------------------------------------------------------
 MMM MM  MM MMM   | Personal Mail   :  tchen@umich.edu                    |
 MMM  MMMM  MMM   | Unimportant Mail:  Antony.Chen@um.cc.umich.edu        | 
 MMM   MM   MMM   |            JESUS CHRIST IS THE LIGHT!                 |
MMMMM      MMMMM  *********************************************************

-----------[000260][next][prev][last][first]----------------------------------------------------
Date:      15 May 1994 10:13:08 GMT
From:      kwia4000@bronto.zrz.TU-Berlin.DE (Manfred Kwiatkowski)
To:        comp.protocols.tcp-ip
Subject:   Re: a SLIP server

In article <L.G.Cuthbert.6.0024B09A@qmw.ac.uk>, L.G.Cuthbert@qmw.ac.uk (Laurie Cuthbert) writes:
> At present we have at QMW a central modem pool with a terminal server that 
> will either allow a standard ASCII serial login or SLIP. The other side is 
> connected to the Campus LAN so that we can then connect anywhere on (or off) 
> site. The problem is that the SLIP implementation is really bad becuase of the 
> way it implements rate adaptation in the direction from the ethernet to the 
> SLIP port.

Hmm, this cannot be the problem, as there is no "rate adaptation" in 
this context. :-) The session related flow-control issues are soley 
up to the two endsystems involved. If they implement it well, you will 
get fair performance over slow links. The only thing a slip server can 
do is to provide a sufficient amount of buffering to support the less 
than average TCP's. One TCP-window is a good figure to start out with.

> The obvious solution is to replace it and this will be done in the medium term.

Not at all! First of all you have to find out where your packets really 
get dropped/delayed/corrupted or lost. Sure, it may be the terminal 
server after all, but if its someting else (more likely), your new box 
will dissappoint you as well.

-- 
Manfred Kwiatkowski         kwiatkowski@zrz.tu-berlin.de

-----------[000261][next][prev][last][first]----------------------------------------------------
Date:      15 May 94 11:35:00 GMT
From:      tony.kempe@digitec.co.za (Tony Kempe)
To:        comp.protocols.tcp-ip
Subject:   Meditech and TCP/IP


BB> I am interested in hearing from anyone using Meditech over TCP/IP
BB> with Telnet. I have some specific and some general questions and can
BB> offer some valuable information that I've discovered during our own
BB> implementation. ANY information or discussion would be welcome. I
BB> can't believe we are alone in doing this...

Currently I am trying to get a WANG system to connect to a Meditech 
system using TCP/IP. The emphasis is on FTP not Telnet, so I can't 
comment on Telnet access, but I hit a big snag with FTP: the Meditech 
system hasn't implemented the ALLOC command and simply responds with an 
error to this command. Not surprisingly, the WANG side aborts the 
transfer. I was able to do FTP, but only when initiated from the 
Meditech system (i.e. GET and PUT). However, was only successful with 
small files. Recently the Meditech system was updated and I will be 
doing more testing this week.

Tony Kempe
* RM 1.3 01840 * 

----
Digitec Online - 20 Nodes  * Johannesburg, South Africa +27-11-476-2008
* Best BBS * Best Sysop *
@digitec.co.za        * RsaNET User Choice Awards 1993 *

-----------[000262][next][prev][last][first]----------------------------------------------------
Date:      15 May 1994 13:07:39 GMT
From:      dswartz@pugsley.osf.org (Dan Swartzendruber)
To:        comp.protocols.tcp-ip
Subject:   Re: Mac to PC TCP-IP link

In article <dafordCpoBn2.2w8@netcom.com> daford@netcom.com (David Ford) writes:
>
>A friend of mine has a Mac and PC and wants to connect them together 
>without going apple-talk like the Coactive Connector.  I was thinking I 
>could put the two together via ethernet.  I was thinking about picking up 
>a tcp-ip adon for both and go.  Any suggestions?  Also, what ftp sites 
>can I find things at to do this?  thanks.

I'm networking an LC-III and a Pentium with no problems.  One
thing to be wary of is the Mac ethernet interface.  Any Mac
which claims to have "builtin ethernet" doesn't really.  At least
not with a compatible interface.  They use some special Apple-only
socket and electronics.  You need a gadget which plugs into the
Mac port and provides 10Base2 (or whatever) on the other.  The ripoff
is that this is almost as expensive as a start-from-scratch
ethernet solution.  If you don't have any such, your avenue depends
on what type of Mac you have.  I have an LC family machine and got
a perfectly adequate ethernet board from Mac Connection (or was it
Mac Warehouse?)  In any event, they have boards from several different
companies.  The board usually includes drivers, although you need
to get MacTCP from somewhere.  Best (cheapest) bet is to buy the
Adam Engst "Internet Starters Kit", take out the floppy with the
included copy of MacTCP, and throw away the book :) [unless you
really want it]


-- 

#include <std_disclaimer.h>

Dan S.

-----------[000263][next][prev][last][first]----------------------------------------------------
Date:      Sun, 15 May 94 23:01:35 PDT
From:      ryker@enterprise
To:        comp.protocols.tcp-ip,alt.winsock,alt.irc.questions
Subject:   need efnet server


Can someone give me an efnet irc server?
ip address please

thanks alot.

-----------[000264][next][prev][last][first]----------------------------------------------------
Date:      Sun, 15 May 1994 16:48:56 GMT
From:      knopf@nisc.jvnc.net (Greg Knopf)
To:        comp.protocols.tcp-ip
Subject:   Re: Where can I get Internet's RFC's?

Joel,

Among other places, you can use anonymous ftp to ftp.jvnc.net
and retrieve the files in /pub/RFC.

The index for all of these RFC's is named rfc-index.txt.

Enjoy,

- Greg
knopf@jvnc.net

In article <ricketts.je.124.2DD3C53F@shocker.ee.twsu.edu>, ricketts.je@shocker.ee.twsu.edu (Ricketts, Joel E.; EE-U) writes:
|> Where would I go about getting Internet's Request for Comment files/
|> documents (RFC's)?  I have been looking for technical information relating
|> to these files, but have had no luck.
|> Joel

-----------[000265][next][prev][last][first]----------------------------------------------------
Date:      Sun, 15 May 1994 17:14:44 GMT
From:      knopf@nisc.jvnc.net (Greg Knopf)
To:        comp.protocols.tcp-ip
Subject:   Re: SUMMARY: FTP problem between Europe and USA

Stephanie,

I'm not sure about this particular link, but in some ftp transfers
I have seen trouble transfering certain data patterns through
multiplexers which convert optical/fiber signals to copper.  The
problem was related to the buildout on the fiber-to-copper
signal.  Adjusting this to 0 allowed the data to pass through.

Since you did not receive anything, though, I would tend to
doubt that this is the problem.  It might be interesting to
to an octal dump of the files in question to look for similarities.

- Greg
knopf@jvnc.net

In article <2r22p9$6qd@sheckley.cnam.fr>, bortz@cnam.cnam.fr (Stephane Bortzmeyer) writes:
|> In article <2qvddb$jbe@sheckley.cnam.fr>, bortz@cnam.cnam.fr (Stephane Bortzmeyer) writes:
|> ...
|> >Well, the problem is apparently solved. At least, it worked this morning 
|> >on all the sites I mentioned.
 ...
|> >Now, the question is: can anynone explain why a f*$#@ problem on a 
|> >line, instead of resulting in loss packets, delays, etc, results in 
|> >impossibility of transferring *some* files and not others? And why
|> >not even one byte is transferred on the offending files?
|> 
|> It's specially strange since during the crisis, ping and traceroute 
|> both reported normal delays and losses. All the other cable problems 
|> I saw (either on Ethernet or on leased lines) were easily reported by 
|> these tools: high percentage of lost packets, incredible delays, etc.
|> 
|> So, nobody gets an idea? (I don't :-( )
|> 
|> Stephane Bortzmeyer           Conservatoire National des Arts et Metiers	
|> bortzmeyer@cnam.fr            Laboratoire d'Informatique
|>                               292, rue Saint-Martin			
|> tel: +33 (1) 40 27 27 31      75141 Paris Cedex 03
|> fax: +33 (1) 40 27 27 72      France	
|> 
|> "C'est la nuit qu'il est beau de croire a la lumiere." E. Rostand
|> 
|> http://web.cnam.fr/personnes/bortzmeyer/home_page.dom
|> 					
|> 	

-----------[000266][next][prev][last][first]----------------------------------------------------
Date:      Mon, 16 May 94 08:55:25 PDT
From:      adar0@routers.com
To:        comp.protocols.tcp-ip
Subject:   Re: Dynamic IP addr & mail



> How is this done? (or where is it explained how to do it)
> I have a dialup IP account with a fixed IP address assigned to my
> system, but I can't receive mail via that account yet.  The way they
> normally solve this at the location I dialup to is to give you another
> login on their machine where you get a shell to read your mail.  I would
> like my mail to be delivered to my own system with SMTP.
> Is there some protocol to trigger sendmail to start sending queued mail?
> 

Rob,
I missed the first part of this thread, however the easiest way to
accomplish downloading mail from your Internet providers unix machine
to your machine (regardless of whether its a PC, MAC or another Unix
machine) is to use the POP3 protocol.  Works like a champ and you
never have to manually log into their machine.
Rich
adar0@routers.com


-----------[000267][next][prev][last][first]----------------------------------------------------
Date:      Sun, 15 May 1994 17:06:19 UNDEFINED
From:      J.M.OConnor@massey.ac.nz (John O'Connor)
To:        comp.os.ms-windows.networking.tcp-ip,comp.protocols.tcp-ip
Subject:   Re: Has anyone seen this symptom with W4WG? (DANGEROUS ICMP behaviour)

In article <2qspgi$8vb@gap.cco.caltech.edu> heathh@wrath.ugcs.caltech.edu (Heath I Hunnicutt) writes:
>From: heathh@wrath.ugcs.caltech.edu (Heath I Hunnicutt)
>Subject: Re: Has anyone seen this symptom with W4WG? (DANGEROUS ICMP behaviour)
>Date: 12 May 1994 08:31:14 GMT
 
>The previous post is complete and total bunk.  
 
>- I am on the network at a campus with a LARGE (>1000) number of UNIX hosts
>using TCP/IP, and many (>50) machines on the same net running WFWg with 
>TCP/IP stacks and NetBEUI.  Some of these machines are running pre-release
>Daytona TCP/IP stacks, to boot.  Not once have I ever heard a single tale
>of such a situation occurring here.  And I would, as I am sort of a Windows
>networking guru around here.  

For the benefit of anyone having this problem, and Windows gurus everywhere, 
this is a reply I sent to Bill Ryder following his original post ...


To: billr@audi.optimation.co.nz (Bill Ryder)
From: J.M.OConnor@massey.ac.nz (John O'Connor)
Subject: Re: Has anyone seen this symptom with W4WG? (DANGEROUS ICMP behaviour)
Date: Thu, 12 May 1994 22:47:59 UNDEFINED
Message-ID: <J.M.OConnor.1.02524B08@massey.ac.nz>
X-status: D

In article <BILLR.94May12154255@audi.optimation.co.nz> billr@audi.optimation.co.nz (Bill
Ryder) writes:
>From: billr@audi.optimation.co.nz (Bill Ryder)
>Subject: Has anyone seen this symptom with W4WG? (DANGEROUS ICMP behaviour)
>Date: Thu, 12 May 1994 03:42:55 GMT

stuff deleted ...

>>
>> Recently, one department installed Windows for Workgroups with the Microsoft
>> IP stacks.  The person at massey was not sure if this software was full
>> release yet.  The WFW machine would send out an ARP, the servers on the
>> network would respond, THEN the WFW machine would send a "ICMP: Host
>> Unreachable" to a random server with a random Hostname.  This server would
>> then shutdown the connections to this random host.

The condition turned out to be particular to the use of NDIS3 drivers with SMC
Elite 16 Ultra (8216) ethernet cards, in conjuntion with Microsoft's TCP/IP
32 betas (1 and 2). Bug reports have been sent to Microsoft and SMC, but no
solution, other than the current workaround of using the NDIS2 driver, has yet
been found.

The following was a previous posting to this newsgroup on the problem :

From: John O'Connor <J.M.OConnor@massey.ac.nz>
Subject: MS-TCPIP beta + SMC Ultra  = ICMP traffic
Date: 9 May 1994 05:08:02 GMT

I've got WFW 3.11 installed on a machine using a SMC Elite 16 Ultra network
card.

With MS-TCP/IP VxD beta 2 installed as the default protocol, and using
the latest NDIS drivers from SMC (ver 3.04), the machine sends
"Destination host unreachable" ICMP packets to other machines on our
network. The destination IP addresses correspond to other hosts on our
net, but range from DOS PC-NFS clients to Unix boxes - seemingly randomly.

When the SMC NDIS driver is configured to use "real" mode instead of
"enhanced", the problem seems to go away. I am certain the setup of the
TCPIP is correct with regard to address, mask, domain, gateway etc. and
the card seems fine using NetBEUI.

I've despatched a bug report to Microsoft, but I'm not convinced that it
is at fault.

Any input would be galdly received...



John O'Connor
Consumer Technology Dept.
Massey University





-----------[000268][next][prev][last][first]----------------------------------------------------
Date:      Mon, 16 May 1994 07:52:50 GMT
From:      richlove@netcom.com (Rich Love)
To:        comp.protocols.tcp-ip
Subject:   Slow Login problem

When logging into a HP host with Telnet TCP/IP, I have heard that slow login
times can be caused by the host trying to mount a floppy drive. This can
happen on a terminal with no floppy or a Macintosh because the host can't
see the floppy. The solution is to remove the "Mount Floppy" command from
the unix login. Has anyone seen this problem on hosts other than HP?
Specifically, I am wondering if this can happen on a Sun host.

-- Rich Love                                     Carnation Software, Inc.
 ________________________________________________________________________
| MacToPic and SBMac - Macintosh to Host connectivity and file transfer  |
| for PICK, uniVerse, unix, System Builder and other host systems.       |
| Viewpoint, Wyse 50, VT101 and Prism emulations included.               |
| Demo available for download at ftp.netcom.com in pub/carnation         |
|                                                                        |
| Phone (206) 333-4288                   Internet:  richlove@netcom.com  |
|________________________________________________________________________|

-----------[000269][next][prev][last][first]----------------------------------------------------
Date:      Mon, 16 May 1994 08:13:09 +0000
From:      proyse@peeras.demon.co.uk (Phil Royse)
To:        comp.protocols.tcp-ip
Cc:        philip@buckley.demon.co.uk
Subject:   Resale of registered IP address space?

Can anybody comment on the likelihood of registered IP address space
being bought and sold (i.e. as an intangible, valuable commodity)?

Putting together the facts that free, unused space is becoming rarer;
the demand for Internet connectivity is increasing; that many 
organisations have vast amounts of unused registered space and are 
in need of cash... you get to speculate on a new market...

Like any diminishing/limited resource, such as unfurnished rented 
accommodation, huge amounts of IP address space could come onto the 
"second user" market if there was any revenue in it....

Also, authorities often are powerless to stop secondary markets
developing... they often just accept the inevitable, but regulate 
it a bit...

I'm old enough to remember the curiosity of seeing the first 
advertisements for people selling vehicle number plates.  It seemed 
wierd at the time (my Dad said it was illegal and wrong and should be 
stopped), but now it's commonplace to see  SUE 123 being sold for 
6,950 UK pounds and  1 AM for 44,000 (yesterday's Sunday Times).  
There is a big market in the UK, which is legitt. and fully supported 
by the govt. registration authority (DVLA).

What would happen if a registered IP address space holder offered
some of its space, e.g. a Class A, to be sold off in Class B and
Class C equivalent chunks?

Would it be against acceptable use?

Is it not covered by any contract law?  What is the contractual
entitlement when you "get an IP address"?  What rights and obligations
do you have?  Do you have any right to sell it?  like a bunch of theatre
tickets? or a Treasury bond?

Would it be better for any secondary "suballocations" to be regulated 
by NIC, RIPE etc. like the service providers are at present?

Would domain names need to be sold as part of the IP address sale?
(i.e. would the buyer of the address space need to be part of the
domain of the reseller)?

What would this resale of IP address space do to Internet routing?

Could address space be leased for an agreed period (e.g six months to
a couple of years)?  Could an org. hedge its bets on the deployment
of IPng by getting revenue now, but reserving the right to reclaim
its address space in a couple of years if IPng was slowed up...?

Is this unthinkable?


Comments, please.

-- 

Phil Royse     Comms Consultant  |  PRA Consulting Ltd.
TUDOR HOUSE                      |
12 Woodside Road, Purley
Surrey  CR8 4LN (UK)  Tel: (+44) 81 645-9868   proyse@peeras.demon.co.uk

-----------[000270][next][prev][last][first]----------------------------------------------------
Date:      16 May 94 15:01:43
From:      drw@runge.mit.edu (Dale R. Worley)
To:        comp.protocols.tcp-ip
Subject:   Re: Resale of registered IP address space?

In article <769075989snz@peeras.demon.co.uk> proyse@peeras.demon.co.uk (Phil Royse) writes:
   Can anybody comment on the likelihood of registered IP address space
   being bought and sold (i.e. as an intangible, valuable commodity)?

   Putting together the facts that free, unused space is becoming rarer;
   the demand for Internet connectivity is increasing; that many 
   organisations have vast amounts of unused registered space and are 
   in need of cash... you get to speculate on a new market...

Last I heard, the present address space was expected to last until
2008, by which time IPng will be up and running.  So one will probably
always be able to get more space for free from the NIC.

*Domain names*, on the other hand, are likely to be worth money,
because they aren't interchangable.

Dale

Dale Worley		Dept. of Math., MIT		drw@math.mit.edu
--
Whipping out one's religion in public should be considered a form of
indecent exposure, and publicly inquiring into someone's religion should
be considered as tactless as publicly inquiring into someone's sex life.

-----------[000271][next][prev][last][first]----------------------------------------------------
Date:      16 May 1994 18:49:32 -0400
From:      alvarez@access.digex.net (Juan Alvarez)
To:        comp.protocols.tcp-ip
Subject:   Looking for software to run TCP/IP on XNS network or vice versa

I am looking for TCP/IP software that will run on PC's running XNS (Ungerman
Bass ethernet cards) - has anybody ever heard of any software that allows
TCP/IP and XNS to run on the same network?  Thanks.

Juan
alvarez@digex.net


-----------[000272][next][prev][last][first]----------------------------------------------------
Date:      16 May 1994 14:38:41 +0100
From:      robjan@rabo.nl (Rob Janssen)
To:        comp.protocols.tcp-ip
Subject:   Re: Dynamic IP addr & mail

In <2qr3gp$gk2@news.CCIT.Arizona.EDU> leonard@telcom.arizona.edu (Aaron Leonard) writes:

>My suggestion, if this is what you're doing, is to stop doing this.
>Instead, assign each dialup client THE SAME address each time it connects.
>(We accomplish this using a Xylogics Annex as our comm server, but I'm
>sure most comm servers can do the same.)  
 
>Each client can be MXed by some mailhub.  When the client connects, that 
>can trigger a dequeuing of the stored messages.

How is this done? (or where is it explained how to do it)
I have a dialup IP account with a fixed IP address assigned to my
system, but I can't receive mail via that account yet.  The way they
normally solve this at the location I dialup to is to give you another
login on their machine where you get a shell to read your mail.  I would
like my mail to be delivered to my own system with SMTP.
Is there some protocol to trigger sendmail to start sending queued mail?

Rob

-----------[000273][next][prev][last][first]----------------------------------------------------
Date:      16 May 1994 14:45:26 +0100
From:      robjan@rabo.nl (Rob Janssen)
To:        comp.protocols.tcp-ip,comp.protocols.tcp-ip.ibmpc
Subject:   Re: TERMINAL SERVER Software for Linux - Where?

In <znr768926930k@leotech.mv.com> epoole@leotech.mv.com (Eric Poole) writes:

>Can anyone direct me to some software that will allow a Linux box to
>make like a terminal server?
 
>Specifically, what I need is hardware/software so that Internoids can
>telnet into here and access our Wildcat BBS ... they would telnet in
>over TCP/IP (well, SLIP), then go out one of the Linux box's serial
>ports to a user port on the BBS.
 
>I plan to use a 386/33 or 386/40 for this, with 8 serial connections
>to the BBS.

Sounds weird and costly to me...
Isn't there any way to run a TCP/IP stack on the BBS system that simulates
some COM ports (INT 14) so that you can telnet to the BBS via an ethernet
connection between the machines?
Such solutions certainly exist for outgoing connections, they are used
by terminal emulators that can access hosts via telnet.  I don't know
if incoming telnet sessions are supported by these products.

Rob

-----------[000274][next][prev][last][first]----------------------------------------------------
Date:      16 May 1994 17:12:16 GMT
From:      ais007@news.salford.ac.uk (Richard Letts)
To:        comp.protocols.tcp-ip
Subject:   Re: BOOTP: where for client?

In article <thinmanCprEAK.4xs@netcom.com> thinman@netcom.com (Technically Sweet) writes:
>From: thinman@netcom.com (Technically Sweet)
>Subject: BOOTP: where for client?
>Date: Fri, 13 May 1994 20:59:07 GMT
>Where would I find sample client code for the BOOTP protocol?
>And, where would I find discussion of what it takes to implement
>this under the BSD flavor of TCP/IP?  (This is for a port of
>BSD TCP/IP into a monolithic real-time kernel, so kernel/user
>space causality & design issues don't matter.)
>
>Please just send mail and I will summarize.
>
>Thanks,
>

check out :

ftp::/ftp.salford.ac.uk/network/misc/bootp.tar.Z
this is a simple bootp server I use on the campus network here.

more interesting:

ftp::/ftp.salford.ac.uk/network/misc/bootp.zip

this contains a program written for LanWorkplace for DOS to spit out a bootp 
packet and decode any responses received. This is probbably what you want 
for the client-side. {written to debug the server}


/RJL
Richard Letts  
-------------------------------------------------------------------------------
Network Manager                               mail:    R.J.Letts@salford.ac.uk
University of Salford                         phone:     +44 61 745 5252
Great Britain                                 fax:       +44 61 745 5888

-----------[000275][next][prev][last][first]----------------------------------------------------
Date:      Mon, 16 May 1994 17:22:59 GMT
From:      dag@tc.fluke.COM (David Gunderson)
To:        comp.protocols.tcp-ip
Subject:   Re: Duplicate IP address detection

The best way I know of detecting if an IP address is active within a 
network (all hosts addressable from a particular host) is to do
a ping on the candidate address. If no host responds, then you know
that that address is currently unused. This is not foolproof because
a host with that address may be just off the net (for example a PC that
happens to be powered down) right now.

--David Gunderson


-----------[000276][next][prev][last][first]----------------------------------------------------
Date:      Mon, 16 May 1994 18:49:16 GMT
From:      b1hg0jv@kato.BELL-ATL.COM (Benash)
To:        comp.protocols.tcp-ip
Subject:   TCP-IP Packet Video?

Hello,

I'm looking for information on packages that will allow me to run video
(motion) on PC's using the IP stack on ethernet LANS. Preferably software
/hardware packages that wouln't dedicate the PC for this use.

Can someone send me in the right direction?

ray@bell-atl.com


-----------[000277][next][prev][last][first]----------------------------------------------------
Date:      16 May 1994 19:22:12 GMT
From:      barmar@think.com (Barry Margolin)
To:        comp.protocols.tcp-ip
Subject:   Re: Gateway trouble, puhleez help!

In article <2qltpn$1eb@xmission.xmission.com> pashdown@xmission.com (Pete Ashdown) writes:
>First off, is there a decent book or document that explains gateways and/or the
>use of traceroute to resolve problems?

"TCP/IP Illustrated" has a chapter on Traceroute.
-- 
Barry Margolin
System Manager, Thinking Machines Corp.

barmar@think.com          {uunet,harvard}!think!barmar

-----------[000278][next][prev][last][first]----------------------------------------------------
Date:      16 May 1994 19:53:19 GMT
From:      barmar@think.com (Barry Margolin)
To:        comp.protocols.tcp-ip
Subject:   Re: Forwarding a socket connection to another machine

In article <TBEL.94May13214336@oahu.cern.ch> tbel@oahu.cern.ch (Tim Bell) writes:
>Is there a way to forward a socket connection to another machine ? I
>was to write a gateway program that will get called when a connection
>is made but then forward on the connection to another machine (it
>won't be the same machine each time).  Since a large amount of data
>will be transferred, I don't want to put a load on the intermediate
>machine so a simple read/write forwarder won't work... in pictures...

There's no built-in mechanism in TCP/IP for this.  Some application
protocols have facilities for this; for instance, FTP has a command that
can be used to tell one server to send a file to another server, rather
that pulling the file to the client and then sending to the second server.

>     A  ------->   B
>			---------> C
>
>     A --------------------------> C

In this case, it seems most straightforward for the A-B connection to start
off with B sending back C's address.  Then A would close the connection
with B and open a new connection with C.
-- 
Barry Margolin
System Manager, Thinking Machines Corp.

barmar@think.com          {uunet,harvard}!think!barmar

-----------[000279][next][prev][last][first]----------------------------------------------------
Date:      Tue, 17 May 94 12:10:00 -0640
From:      john.will@dscmail.com (John Will)
To:        comp.protocols.tcp-ip
Subject:   Re: Resale of registered

D >*Domain names*, on the other hand, are likely to be worth money,
D >because they aren't interchangable.

Why would a domain name be worth anything?  I can register as many 
domain names as I can invent, doesn't seem to be a commodity...

-----------[000280][next][prev][last][first]----------------------------------------------------
Date:      17 May 1994 01:51:25 GMT
From:      prep@yarrow.wt.uwa.edu.au (Paul Repacholi)
To:        comp.protocols.tcp-ip
Subject:   Fuzzball code


Does anyone still have a copy of the FUZZBALL sources arround
they could send me a copy of?

TNX
--
~Paul
                                                       +61 (09) 257-1001
prep@yarrow.wt.uwa.edu.au ( preferred )                1 Crescent Rd,
zrepachol@cc.curtin.edu.au                             Kalamunda,
                                                       West Aust    6076


-----------[000281][next][prev][last][first]----------------------------------------------------
Date:      Tue, 17 May 1994 02:35:31 GMT
From:      c352c03@hawk.depaul.edu (Alex L. Kudlov)
To:        comp.protocols.tcp-ip
Subject:   Re: IP Addresses and the Internet?

>
>My own quick analysis of RFC-1597 led me to find what I think is a new
>and significant potential security attack arising when a typical
>firewall is used with networks configured as RFC-1597 suggests.  
>I haven't confirmed the existence of the attack yet myself, so I
>
  What exactly is RFC?  Is that some sort of a text file?  Where can
I obtain one?




-----------[000282][next][prev][last][first]----------------------------------------------------
Date:      Tue, 17 May 1994 04:57:20 GMT
From:      ericd@gaudi.csufresno.edu (Eric Douglas)
To:        comp.protocols.tcp-ip
Subject:   Re: Duplicate IP address detection

Re: Finding duplicate IP addresses on LAN. If your machines
which possibly have the same IP support a full TCP stack, 
it seems possible that opening a TCP connection to an arbritrary
port (there must be a server listening on it) then checking the
number of SYN/ACK replies would give you a fairly accurate 
number. I would think that both machines would try to negotiate
the connection. However, this assumes that you have access to
a hardware level packet reading mechanism, such as /dev/nit.
You couldn't get at this through the normal TCP stack...

--eric



-- 
/****************************************************************************
* Eric W. Douglas           Internet: eric_douglas@csufresno.edu
* CCMS/CSci, California     Voice: +1 209 278 3923 or +1 209 897 4556
* State University Fresno   Logic joke: if p->q and not q then not p, NOT!

-----------[000283][next][prev][last][first]----------------------------------------------------
Date:      17 May 94 17:21:02 -0500
From:      harvey@indyvax.iupui.edu (James Harvey)
To:        comp.protocols.tcp-ip
Subject:   Re: Duplicate IP address detection

In article <CpyE7E.JA3@ilx.com>, skov@ilx.com (John Skovron) writes:

	[stuff deleted]
>
> There's another kind of problem that I've run into.  Lan administrators
> in the DOS world are used to working with Novell and NetBios protocols
> which are not so picky about address administration as is IP.  They have
> a tendancy to create new systems by copying existing systems, so if you
> have provided a system with the IP address configured on the hard drive
> (via ifcust or net.cfg or protocol.ini), the LAN administrator will
> soon copy your system with its IP address intact, creating duplicate IP
> addresses on that subnet.

This can be a continual problem on a large net.  For example here at IUPUI
recently we had at least 105 duplicates out of 4,251 addresses in use.
You can obtain this kind of information by periodically polling the ARP tables
on your routers.  You may be better off giving your LAN admins something they
can copy blindly and then using RARP or BOOTP to assign all addresses.

>                            One way around this is to write a small
> utility program that uses the low-level PKTD, ODI, or NDIS interface to
> send out an ARP message.  The ARP message contains 0.0.0.0 as its source
> protocol address, and has the desired IP address for this system (that
> is, the IP address which we are checking for duplicates) as the target
> protocol address.  (Otherwise the ARP packet looks very much like a
> standard IP ARP.)  If there is no response within some timeout/retry,
> then the address is not duplicated and can be used.  If there is a
> response, then the address is duplicated and should not be reused.

Should broken proxy ARP implementations (i.e. replies incorrectly sent to the
interface from which they were "learnt") be checked for?  How common are they?

All the TCP/IP stacks I've seen already do this when bringing up an interface
on a broadcast LAN.  Bad Things could happen if some bonehead could bring up a
PC with a router's IP address.
--
James Harvey   harvey@iupui.edu   IUPUI OIT Networks and Systems Tech Support
Disclaimer:  These are my own opinions.  I do not speak for Indiana University.

-----------[000284][next][prev][last][first]----------------------------------------------------
Date:      Tue, 17 May 1994 09:33:39 GMT
From:      wiersma@seri.philips.nl (Dick Wiersma)
To:        comp.protocols.tcp-ip
Subject:   ftp patch to support struct option

Questions:
1.  Has anybody done something in this area on Sun's ftp?
2.  Is there a PD ftp version that supports 'struct'?
3.  Any other useful hints?

Thanks.
--
-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-
Dick Wiersma                                          wiersma@seri.philips.nl
Tel.: (040-7)86812                                      SERI: wiersma@phcoms0
Fax:  (040-7)82900                           Philips C&P MPN services, VA-129

-----------[000285][next][prev][last][first]----------------------------------------------------
Date:      Tue, 17 May 1994 15:40:25 GMT
From:      skov@ilx.com (John Skovron)
To:        comp.protocols.tcp-ip
Subject:   Re: Duplicate IP address detection

In article <CpwoAG.8G1@tc.fluke.com>, David Gunderson <dag@tc.fluke.COM> wrote:
>The best way I know of detecting if an IP address is active within a 
>network (all hosts addressable from a particular host) is to do
>a ping on the candidate address. If no host responds, then you know
>that that address is currently unused. This is not foolproof because
>a host with that address may be just off the net (for example a PC that
>happens to be powered down) right now.
>
>--David Gunderson
>

This method will work, _provided_ that you already have IP up and running
on a machine with a unique (non-duplicated) IP address.  I missed the
original post, so I don't know if that's the situation that the original
poster was working from.  Also, this approach isn't guaranteed to  help
if there are already duplicate IP addresses up on the net, since the
presence of duplicate addresses will confuse some implementations and
subnets so badly that they cannot properly respond to a ping.

There's another kind of problem that I've run into.  Lan administrators
in the DOS world are used to working with Novell and NetBios protocols
which are not so picky about address administration as is IP.  They have
a tendancy to create new systems by copying existing systems, so if you
have provided a system with the IP address configured on the hard drive
(via ifcust or net.cfg or protocol.ini), the LAN administrator will 
soon copy your system with its IP address intact, creating duplicate IP
addresses on that subnet.  One way around this is to write a small
utility program that uses the low-level PKTD, ODI, or NDIS interface to
send out an ARP message.  The ARP message contains 0.0.0.0 as its source
protocol address, and has the desired IP address for this system (that
is, the IP address which we are checking for duplicates) as the target
protocol address.  (Otherwise the ARP packet looks very much like a
standard IP ARP.)  If there is no response within some timeout/retry,
then the address is not duplicated and can be used.  If there is a
response, then the address is duplicated and should not be reused.

I've verified that this approach works with several different stacks
and with several different IP implementations on the local subnet.  I
think that the new DHCP RFCs also use a similar approach.  These RFCs
(1541, 1542, et al) are probably the best reference for how to do this
with upwards compatibility.  I imagine that a number of vendors (Sun
and Microsoft and others) will use this approach as they implement
DHCP.


-- 
-----------------------------------------------------------------------------
John Skovron
ILX Systems Inc.
skov@ilx.com

-----------[000286][next][prev][last][first]----------------------------------------------------
Date:      18 May 1994 01:01:46 -0500
From:      Samuel.Dennis@f1.n700.z6.ftn.air.org (Samuel Dennis)
To:        comp.protocols.tcp-ip
Subject:   SCO IPX/SPX

Thank you for your answer about SCO IPX/SPX. Now my login name change from
"Samuel Leo" to "Samuel Dennis". We were going to make a program to access 
Netware with SCO IPX/SPX just before your answer arrived. Do you think it
is still necessary to make the program? And when will your product come into
market? 
I've tried to connect you by E-mail but my machine did not recognize your
address "sco.com". It told me "unknown host".
Can you tell me if SCO has product in multimedia area? We and many of our
users want to use sound blaster and video blaster in SCO ODT.
My E-mail address is sdlab@iohk.com.

Best regards


... Maximus 2.00
 # Origin: TAIC Maximus-CBCS - Home of HK PC User Group (6:700/1)

-----------[000287][next][prev][last][first]----------------------------------------------------
Date:      Tue, 17 May 1994 16:55:13 GMT
From:      mgalloway@worldbank.org (Monica Galloway)
To:        comp.protocols.tcp-ip
Subject:   Best Dial-In Options?

We are developing a client-server system on Macintosh clients and Sun
servers (SPARC-10s) which commmunicate using TCP/IP protocols over an
internal IP network.  The final system will be deployed on both Mac and
Windows clients, both internal and external.  The external clients will be
accessing the Sun servers via dial-up connections; dial-ups can be from
portables.  I've been hearing lots of stuff about different options and am
searching for the best solution...with security addressed.

Can anyone give me some tips?  What, exactly, do you recommend in terms of
software for our clients and servers to make it all work? 

Please email replies.  Thanks.
Monica Galloway
mgalloway@worldbank.org

-----------[000288][next][prev][last][first]----------------------------------------------------
Date:      17 May 1994 23:59:44 -0400
From:      reh@cs.umd.edu (Richard Huddleston)
To:        comp.protocols.tcp-ip
Subject:   Re: routed QUESTION

quang@pairgain.com (Quang Bui) writes:

>I have problem getting pass the gateway to get out to the Internet.
>Here is the network configuration. On my main subnet which connects to
>the Internet I have no problem getting to the Internet from any node,
>but from any subnet which connect to the main net, I cannot get out
>to the Internet at all, to get around this I have to rlogin to
>the main net. Is this a problem with routing, DNS, proxy server, etc...

Hard to tell from here ;) -- but it sounds like the router(?) that's
in between all of your subnets doesn't have a (correct) default route 
entry pointing at the Internet gateway IP address.  Do you have two
routers on either end of your "main" subnet -- and does one of those
routers connect to the 'net ?

If it's not a simple routing problem, then more information is needed
in order to be helpful.





>
>


-- 
ObIdentifier:		Richard Huddleston 
ObDisclaimer:		Not an official spokesperson for HCIL or UMD 

ObFriendlyReminder:	Power is dangerous, and must be handled prudently. 

-----------[000289][next][prev][last][first]----------------------------------------------------
Date:      Tue, 17 May 1994 18:48:54 GMT
From:      venu@martha.utcc.utk.edu (Nair Venugopal)
To:        comp.protocols.tcp-ip
Subject:   Steven Bellovin's paper on security(1989)

Hi, 
I was trying to get access to S.M. Bellovin's paper titled
"Security Problems in the TCP/IP Protocol Suite" published
in Computer Communication Review (Vol 19, No: 2; April 1989).

Our library does not have it. I would be grateful if some one
could email me either the name of an  ftp site having it or
a postscript version of the paper.

Thanks

Venu
(venu@utkux.utcc.utk.edu)


-----------[000290][next][prev][last][first]----------------------------------------------------
Date:      17 May 1994 19:50:48 GMT
From:      edm@harpo.dev.uga.edu (Ed Maioriello)
To:        comp.protocols.tcp-ip
Subject:   Re: Duplicate IP address detection

In article <1994May13.230125.10719@dms.agames.com>,
Mike Albaugh <albaugh@agames.agames.com> wrote:
>
>	He did say "On a LAN" which would possibly lead to strategies
>using ARP. I know it's not a general solution, but it can help. Now if we
>can only get rid of the printer-boxes that respond to random ARPS :-)
>

On a semi-serious note, given that most of the conflicts I found occur
in an office where someone has copied someone else's software - config
files and all, I like to use the "Oh Shit" method.  Basically this
involves one continuing to use the address in conflict while someone
walks up and down the halls listening for the user who says "Oh Shit!".

Otherwise, only careful tracking of nic addresses can keep one out of
trouble.

Ed Maioriello                                         edm@eris.ucns.uga.edu
University Computing & Networking Services            edm@harpo.dev.uga.edu
University of Georgia  ----------------------------------------------------
Athens, Ga. 30602      | First Rule of Troubleshooting: 
(706)-542-6468         |                     If it don't work - plug it in! 


-----------[000291][next][prev][last][first]----------------------------------------------------
Date:      Tue, 17 May 1994 20:23:39 GMT
From:      prime@herring.dr.att.com (Anthony R. Davis)
To:        comp.protocols.tcp-ip
Subject:   What are RFCs

Alex writes:
>What exactly is RFC?  Is that some sort of a text file?  Where can
>I obtain one?


Alex,
RFC is an acronym for Request For Comment. These are a series of documents
that describe various technical aspects of the Internet. They define the
protocols used in the IP Internet.  Some are only suggestions, some are
even jokes, and others are published standards.  Several sites in the
Internet store RFCs and make them available for anonymous ftp from the
following servers:

	FTP.NISC.SRI.COM 
	NIC.DDN.MIL
	nic.cerf.net

There are many others, but this is a good start! Hope this helps!


Tony

It is not "Many are called, but few are chosen", but rather "ALL ARE CALLED,
BUT FEW CHOOSE TO LISTEN".....Think about it!!

                ******** 
             *----*********
            *-------********
           *---------******** 
           *--------********* 
           *-------*********
            *----**********
               *********

	         TERA-X
..........................................................................
Anthony Davis
AT&T Bell Laboratories
Denver,  Colorado
net:   prime@dr.ATT.COM

-----------[000292][next][prev][last][first]----------------------------------------------------
Date:      Wed, 18 May 1994 07:48:31 -0500
From:      lgrande@hr.house.gov (Len Grande)
To:        comp.protocols.tcp-ip
Subject:   Re: TCP-IP Packet Video?

In article <CpwsA4.6L0@kato.BELL-ATL.COM>, b1hg0jv@kato.BELL-ATL.COM
(Benash) wrote:

> Hello,
> 
> I'm looking for information on packages that will allow me to run video
> (motion) on PC's using the IP stack on ethernet LANS. Preferably software
> /hardware packages that wouln't dedicate the PC for this use.
> 
> Can someone send me in the right direction?
> 
> ray@bell-atl.com


Cornell University has developed a packet video product called CU-SeeMe
that runs on the TCP/IP stack.  I believe they have a version for both Macs
and PCs.  From what I understand, this product provides grayscale video
images and does not provide an audio connection at this time. However, its
freeware.

There are also several commercial vendors that are offering packet video
products running on TCP/IP.  InVision Systems Corp. of Vienna, VA is one
that I'm aware of.


Len 

-- 
Len Grande
Network Systems Engineer
House Information Systems
lgrande@hr.house.gov
"The views expressed are my own and are not those of my employer."

-----------[000293][next][prev][last][first]----------------------------------------------------
Date:      18 May 1994 14:48:31 -0700
From:      lelina@kaiwan.com (Rico Lelina)
To:        comp.protocols.tcp-ip
Subject:   Re: TCP-IP Packet Video?

In article <lgrande-180594074831@lensmac.house.gov>, Len Grande wrote:
> In article <CpwsA4.6L0@kato.BELL-ATL.COM>, b1hg0jv@kato.BELL-ATL.COM
> (Benash) wrote:
>
> > Hello,
> >
> > I'm looking for information on packages that will allow me to run video
> > (motion) on PC's using the IP stack on ethernet LANS. Preferably software
> > /hardware packages that wouln't dedicate the PC for this use.
> >
> > Can someone send me in the right direction?
> >
> > ray@bell-atl.com
>
>
> Cornell University has developed a packet video product called CU-SeeMe
> that runs on the TCP/IP stack.  I believe they have a version for both Macs
> and PCs.  From what I understand, this product provides grayscale video
> images and does not provide an audio connection at this time. However, its
> freeware.
>
> There are also several commercial vendors that are offering packet video
> products running on TCP/IP.  InVision Systems Corp. of Vienna, VA is one
> that I'm aware of.

Abacus Programming Corporation of Canoga Park, CA is another.

-----------[000294][next][prev][last][first]----------------------------------------------------
Date:      Wed, 18 May 1994 04:19:01 GMT
From:      ericd@gaudi.csufresno.edu (Eric Douglas)
To:        comp.protocols.tcp-ip
Subject:   Re: Steven Bellovin's paper on security(1989)

In article <1994May17.184854.22633@martha.utcc.utk.edu>,
Nair Venugopal <venu@martha.utcc.utk.edu> wrote:
>Hi, 
>I was trying to get access to S.M. Bellovin's paper titled
>"Security Problems in the TCP/IP Protocol Suite" published
>in Computer Communication Review (Vol 19, No: 2; April 1989).

You can probably find it by anonymous ftp at:

	research.att.com:/dist/smb

along with the rest of his papers.

--eric
-- 
/****************************************************************************
* Eric W. Douglas           Internet: eric_douglas@csufresno.edu
* CCMS/CSci, California     Voice: +1 209 278 3923 or +1 209 897 4556
* State University Fresno   Logic joke: if p->q and not q then not p, NOT!

-----------[000295][next][prev][last][first]----------------------------------------------------
Date:      Wed, 18 May 1994 04:59:14 GMT
From:      nathan@seldon.foundation.tricon.com (Number 6)
To:        comp.protocols.tcp-ip
Subject:   Re: Encyption of data between nodes across the net

In article <2qcs1e$8bc@lykos.netpart.com> phil@lykos.netpart.com (Phil Trubey) writes:
>In article <DRW.94Apr24142356@nevanlinna.mit.edu>,
>Dale R. Worley <drw@nevanlinna.mit.edu> wrote:
>>In a perfect world, the routers that gateway from Office A to the
>>outside world and from Office B to the outside world would
>>encrypt/decrypt all outgoing packets that were to/from the other
>>office.  But I doubt that such a router exists.
>
>From the Morningstar Web Server (http://www.morningstar.com/):
>
>Our Internet connectivity products provide allimportant
>security features, such as a robust firewall packet filter
>(keeps prying eyes off your LAN), link peer authentication
>(verify the identity of the host on the other end of the wire),
>selective gateway encryption (ensures the privacy of all
>communications that traverse the Internet between your LAN and
>your branch offices and business partners), and other useful
>tools.
>
>If I'm not mistaken, that last feature refers to Morningstar's 
>ExpressRouter which does selective TCP level encryption between 
>compatible routers.
>

Sorry to come in so late on this...haven't read this group in
ages.

I just got my Morningstar Express and it does indeed support 
encryption.

Basically, here is how it works:

On one Express router, you set up a file called "Keys" in its flash
memory.  On the other one you do the same.  The file contains
an IP address (your end) and another address (the other end) and
a key.  You may either specify networks or individual routers/ip
addresses in this file.  Anything going to an address in the Keys
file will be encrypted, anything NOT going to an address in the
Keys file will NOT be encrypted.  An excellent solution that 
escaped my sometimes feeble brain when I first heard about 
encryption!  ;)

I haven't tried it on my Express router (I *just* got it about
four hours ago).  I can say, though, I like the Express.  It was
$2K and has one ethernet and 3 serial ports, one supports up to
T1 and the others support 0 to 115Kbps (all available simultaneously).
Took me about 1/2 hour to get it going, since I could basically copy
all of my current Morningstar PPP on my host settings over to the
Express.

As usual:  not affiliated with Morningstar, but have been a customer
for nearly two years now, mostly happy (very happy with the Express
so far!)  One thing I especially like about the express is that it
feels like a normal Unix box...how they fit a unix kernel into
flash memory is beyond me!  ;)  (just kidding).  It has ls, ftpd,
netstat, ifconfig, rm, ps, kill, etc. so it feels exactly like
setting up TCP/IP on a Unix host.  Unlike some other routers with
crazy command interfaces.

Hope this helps someone.




***********************************************************************
Nathan D. Lane, VP Triicon Systems. Lompoc, CA   
NaN != 6, 6 == 1.  I am not a number, I am a free list!
I'm a programmer  my computers are more valuable than my cars.

-----------[000296][next][prev][last][first]----------------------------------------------------
Date:      17 May 94 15:43:34 +1000
From:      u1066579@csdvax.csd.unsw.edu.au
To:        comp.protocols.tcp-ip
Subject:   http??

Hi,
   Can someone please tell me what http is?  Is it an alternative file transfer
method?
  Please E-mail if you can help.
Best Wishes,
Henry.


-----------[000297][next][prev][last][first]----------------------------------------------------
Date:      Wed, 18 May 1994 09:17:47 GMT
From:      perretc@eig.unige.ch (Perret Cedric)
To:        comp.protocols.tcp-ip
Subject:   HELP PPP PC-SUN

Hello, 
Can someone tell me if it is possible to use the driver etherPPP to comunicate with a SUN
which use the internal PPP of Solaris 2.3?

Thanks to E-mail ME
				Perret Cedric.



-----------[000298][next][prev][last][first]----------------------------------------------------
Date:      18 May 1994 10:38:40 GMT
From:      tin@enst.fr (News Reader)
To:        comp.protocols.tcp-ip
Subject:   tcdump on Solaris 2.x

Hello folks.

Is anyone aware of a port of TCPDUMP under Solaris 2.x ???
I would like to use it on a ATM testbed to analyze the tcp/ip
implementation differences between SunOs and Solaris 2.x.

Replies to Olivier.Danthine@ici.der.edf.fr since I can't access
easily to this newsgroup...

Thanks for your help !

O.D.


-----------[000299][next][prev][last][first]----------------------------------------------------
Date:      18 May 1994 11:13:18 GMT
From:      root@jedi.edb.tih.no (Ragnvald T. Blindheim)
To:        comp.protocols.tcp-ip
Subject:   Re: tcdump on Solaris 2.x

News Reader (tin@enst.fr) wrote:
: Is anyone aware of a port of TCPDUMP under Solaris 2.x ???

Hope you got my mail.  

Anyway, I followup here since I know several people would want this
info too.
I got this info from Ng Pheng Siong ngps@np.ac.sg.
At ftp.cs.vu.nl, /pub/leendert/tcpdump.tar.gz is a port of tcpdump
that uses dlpi.  It should work on Solaris.
As far as I know Leendert is the guy that ported it.


--
Ragnvald T. Blindheim           ragnvald@edb.tih.no
Utleirveien 6
7033 Trondheim
Norway

-----------[000300][next][prev][last][first]----------------------------------------------------
Date:      18 May 1994 13:37:07 GMT
From:      john@iastate.edu (John Hascall)
To:        comp.protocols.tcp-ip
Subject:   Re: Resale of registered

John Will <john.will@dscmail.com> wrote:
}D >*Domain names*, on the other hand, are likely to be worth money,
}D >because they aren't interchangable.
}
}Why would a domain name be worth anything?  I can register as many 
}domain names as I can invent, doesn't seem to be a commodity...

make_money_fast()
{
	foreach company in fortune(500) do
              if (!whois(company+".com")) register(company+".com")
        done
}
-- 
John Hascall                   ``An ill-chosen word is the fool's messenger.''
Systems Software Engineer
Project Vincent
Iowa State University Computation Center  +  Ames, IA  50011  +  515/294-9551

-----------[000301][next][prev][last][first]----------------------------------------------------
Date:      Wed, 18 May 1994 14:16:58 GMT
From:      leendert@cs.vu.nl (Leendert van Doorn)
To:        comp.protocols.tcp-ip
Subject:   Re: tcdump on Solaris 2.x

root@jedi.edb.tih.no (Ragnvald T. Blindheim) writes:

# Anyway, I followup here since I know several people would want this
# info too.
# I got this info from Ng Pheng Siong ngps@np.ac.sg.
# At ftp.cs.vu.nl, /pub/leendert/tcpdump.tar.gz is a port of tcpdump
# that uses dlpi.  It should work on Solaris.
# As far as I know Leendert is the guy that ported it.

Nope, it doesn't work on Solaris, only on SVR4.0 for [34]86 Intel
architectures. My, quick and dirty port, uses DLPI I. I thought
about rewriting it for DLPI II so that it could run on Solaris
but never got around to that.

	Leendert
-- 
Leendert van Doorn 			   		<leendert@cs.vu.nl>
Vrije Universiteit / Dept. of Math. & Comp. Sci.	+31 20 5484477
Amoeba project / De Boelelaan 1081A
1081 HV Amsterdam / The Netherlands

-----------[000302][next][prev][last][first]----------------------------------------------------
Date:      Wed, 18 May 1994 14:18:30 GMT
From:      bruceo@loki.isc-br.com (Bruce Oscarson)
To:        comp.protocols.tcp-ip
Subject:   HELP: t_bind assigning a different address


I am having a problem when using t_bind with TCP/IP, and hope that 
someone might have some recommendations on how to get around the 
problem.

I have a simple client (NT or Unix 5.4) and server (Unix 5.4). The 
Unix applications are written using TLI, NT applications with
winsock.  If the server respawns while the client is connected,
the client is not able to reconnect.  The server must be
restarted again before the client is able to see it.

Using netstat, I monitored the TCP connections.  When all is well, 
the server's local address was listed as host.service, and the 
client showed that it was connected to host.service.  When the 
server goes down, the client connection state goes to TIME_WAIT 
(we are T_COTS_ORD).  It the server restarts during the period 
when the client is in TIME_WAIT state, the server's local address 
becomes something like host.1112, and clients are unable to 
connect.

With the debugger, we looked at the t_bind structure.  The request 
structure contained the address that included the socket number (8926)
for our service and the IP address of the host.  The response 
t_bind structure had a socket number around 1100.

We tried setting the qlen element to 0, but did not detect any 
improvement in the situation.

It is important that we are able to respawn from a failure and 
provide timely service.  The TIME_WAIT state lasts for almost 1 
minute, which is too long to delay before restarting the service.

Being rather new to TCP/IP, I am not certain if I have left out 
any important details.  I would appreciate your suggestions on how 
to avoid the 1 minute delay and how to avoid having t_bind assign 
a different address.

Thanks!

bruceo@mail.isc-br.com
--

Bruce Oscarson          | bruceo@mail.spk.olivetti.com
Olivetti North America  |
N7RWO                   | ma-bell  (509) 927-5437

-----------[000303][next][prev][last][first]----------------------------------------------------
Date:      18 May 94 14:30:15 GMT
From:      chrisc@fir.canberra.edu.au (Chris Chlap)
To:        comp.protocols.tcp-ip
Subject:   Re: how to RARP?

In <BRADLEY.94May13140518@jhawk.wg.com> bradley@wg.com (Doug Bradley) writes:


>could any kind soul out there please point me to some reference
>(or better yet example source) on how to make a rarp query?
>rfc903 was not much help.
Try the code from Comer & Stevens: Internetworking with TCP/IP Vol. 2.
Note that the code you need is the function called rarpsend, which is
NOT listed in the book, but is available via anonymous FTP from
arthur.cs.purdue.edu in /pub/dls/xinu.7.9.tar.Z (or similar). A PC
version is available from csc.canberra.edu.au in /pub/ise/xinu as
xinu79.exe (a self-extracting archive).
Chris Chlap
University of Canberra, Australia


-----------[000304][next][prev][last][first]----------------------------------------------------
Date:      18 May 1994 13:39:19 +0100
From:      js@calvin.lif.icnet.uk ( BIU)
To:        comp.protocols.tcp-ip
Subject:   Is there a FAQ for this newsgroup?

Is there a FAQ for this newsgroup, and if so, where?
-- 
       Jack               js@biu.icnet.uk
                         
If you only have a hammer, you tend to see every problem as a nail.
                -- Maslow

-----------[000305][next][prev][last][first]----------------------------------------------------
Date:      Wed, 18 May 1994 16:15:46 GMT
From:      claude@genethon.genethon.fr (Claude Scarpelli)
To:        comp.protocols.appletalk,comp.protocols.tcp-ip,comp.sys.mac.comm,fr.network.divers
Subject:   MacTCP 2.0.4 & BOOTP RFC 1048 & DNS

I try to configure MacTCP 2.0.4 with BOOTP. BootP server is running on
SunOS4.1.3, and MacTCP is running on a Quadra 650 (System 7.1 french).

On the MacTCP side, I select "server".

On BootP server I filled the following field:

mac219.genethon.fr:\
	ht=ethernet:\
	sm=255.255.255.128:\
	gw=192.70.43.254:\
	ds=192.70.43.254:\
	ha=0800073cf261:\
	ip=192.70.43.219:\
	hn:

I expect this to be:
Hardware Type 		Ethernet
Subnet Masks		Class C with one bit
Gateway			192.70.43.254
Domain Name Server	192.70.43.254
Hardware Address	0800073cf261 (the one of the Mac Quadra)
IP adress		192.70.43.219
Send Hostname		YES



When I reboot the Mac, it sends a BootP request. My BootP server answers :
halogene# /usr/local/etc/bootpd -s -d -d -d -d -d -d -d -d -d -d -d -d -d -d 
bootpd: reading new "/etc/bootptab"
bootpd: read 7 entries from "/etc/bootptab"
bootpd: request from hardware address 0800073CF261
bootpd: found 192.70.43.219 mac219.genethon.fr
bootpd: vendor magic field is 99.130.83.99
bootpd: sending RFC1048-style reply


The Mac record its IP address, the gateway address and the subnet mask
correctly. But it does not get neither the domain name nor the name
server adress.

Can someone tell me how to tell the Mac its domain name and name server
address ?


-- 
------------------------------------------------------------------------------
Claude Scarpelli                        Internet : claude@genethon.fr
Centre de Recherche sur le Génome Humain

-----------[000306][next][prev][last][first]----------------------------------------------------
Date:      18 May 1994 17:07:08 GMT
From:      tkaras@mason1.gmu.edu (THEODORE R. KARAS)
To:        comp.programming,comp.client-server,comp.protocols.tcp-ip,comp.graphics
Subject:   Re: Client  -  Server Development for Graphic Applications

In article <2r03i6$ael@nyx10.cs.du.edu> sbkumar@nyx10.cs.du.edu (suresh kumar) writes:
>
>
>Hi all:
>
>I need to develop a client - server application. The server is going
>to be a statistical or some database system which takes in some data
>and plots some graphs and gives it back to the client. It will be
>running on an Unix machine. The client will be either on a pc or some
>Unix workstation. The easiest thing would be to transfer the data from
>the server to the client nad plot it on the client's machine, but i dont
>want the client (the end user) to buy some software to do it, as it will
>add the cost.
>Can anyone throw some light as how i can go about it -- the easiest
>and fastest way to transfer the plotted file (a GIF file) from the 
>server to the client. I am not sure with the limitations of using
>sockets. Any advice will be highly appreciated.
>
>Thanks
>suresh
>

Note, in my reply I am reversing client-server terminology, since its
reversed in X terminology.

I am going to be doing something similar to this. My question is how are
you going to display graphics on a pc from a UNIX client? Are you going
to write a network application using sockets? One alternative would be
to use NT, but I'm not sure. I think there are three ways to do this -

(1) Send data from client to server, have the pc take the data and then
    do the graphing. 
(2) Convert data on the client side to some kind of graphical format
    (i.e. circles, points, etc.), send the graphical objects to the
    server, the server then displays the graphical objects.
(3) The simplest choice is to do all the graphical functions in UNIX,
    I assume X (?), and then just send pixmap data over the network.


(3) is slow, but very simple. If you send a complete screen, 
~ 1000 X 1000 pixels, with each pixel ~2 bytes, then that's 2 Megs
in one draw. That's alot of data, assuming you transfer the whole
screen. If its not an interactive visualization, that's how I would do.


I don't know. I wish I understood more. If you have any info., please
send it to me.


Ted Karas



-----------[000307][next][prev][last][first]----------------------------------------------------
Date:      Wed, 18 May 1994 22:36:56 UNDEFINED
From:      jlewis@inorganic5.chem.ufl.edu (Jon Lewis)
To:        comp.protocols.tcp-ip,comp.protocols.tcp-ip.ibmpc
Subject:   Re: KA9Q with dial-in slip.

In article <May.18.21.38.06.1994.21147@remus.rutgers.edu> rtkao@remus.rutgers.edu (Richard Kao) writes:

>I have been trying to setup a dialin slip server and I have it working
>after some work.  However, I had to use a second IP address in our
>existing network which I didn't register with our network
>administrator.  I would like to setup the slip server as follows...
 
> ------------------                            -----------------
>| Existing Network |      ---------------     | New Domain for  |
>|                  |-----| KA9Q SLIP     |----| SLIP Connetions |
>| IP 141.202...    |     | Server/Router |    | IP 141.200...   |
> ------------------       ---------------      -----------------
 
>I wan't the SLIP connections to have a completly different address.
>This way I don't have to register it with our network people.  Is it
>possible to do this with out changing anything in our existing
>network.  Is it possible to telnet to IP address 141.202.200.1 from a
>slip connection IP address 141.200.200.1.  Does anyone have any ideas?

Does this other domain (141.200) belong to you?  Or are you planning to just 
steal an address from someone else's domain?




|------------------------------------------------------------------|
| jlewis@inorganic5.chem.ufl.edu   If the first address bounces    |
|                                  let me know, and resend to the  |
| jlewis@chem.ufl.edu              second address.                 |
|                                  If the second address bounces   |
|                                  something's seriously wrong.    |
|                                                                  |
|                   Mime attachments are OK                        |
|__________________________________________________________________|



-----------[000308][next][prev][last][first]----------------------------------------------------
Date:      18 May 1994 16:24:37 +0100
From:      js@calvin.lif.icnet.uk ( BIU)
To:        comp.protocols.tcp-ip
Subject:   Re: Resale of registered

In article <2rd5m3$qib@news.iastate.edu> john@iastate.edu (John Hascall) writes:
>John Will <john.will@dscmail.com> wrote:
>}Why would a domain name be worth anything?  I can register as many 
>}domain names as I can invent, doesn't seem to be a commodity...
>
>make_money_fast()
>{
>	foreach company in fortune(500) do
>              if (!whois(company+".com")) register(company+".com")
>        done
>}
Yeah, I had this idea a few months back. I guess someone
else is going to implement it before me. Sigh.
Still, I think I'll get fxck.com first.
-- 
       Jack               js@biu.icnet.uk
                         
If you only have a hammer, you tend to see every problem as a nail.
                -- Maslow

-----------[000309][next][prev][last][first]----------------------------------------------------
Date:      Wed, 18 May 1994 19:41:39 GMT
From:      epedegu@inf.utfsm.cl (Gabriel Epeldegui)
To:        comp.protocols.tcp-ip,alt.winsock,alt.irc.questions
Subject:   Re: need efnet server

ryker@enterprise wrote:

: Can someone give me an efnet irc server?
: ip address please

Yep

and is available via anonymous ftp from h.ece.uiuc.edu as
/irc/servers.940401.

-----< Africa >-------------------------------------------------------------

-----< Asia >---------------------------------------------------------------

Israel

  birisc.cs.biu.ac.il     2.8.16     Bar-Ilan University
  irc.tau.ac.il           2.8.16     Tel-Aviv University
  irc.technion.ac.il      2.8.16     Technion, Haifa

Japan

  dec303.aist-nara.ac.jp  2.8.16     NarA Institute of Science and Technology
  dec504.aist-nara.ac.jp  2.8.16     NAra Institute of Science and Technology
  wsclark.huie.hokudai.ac.jp 2.8.16  Hokkaido University,Sapporo
  bohemia.jaist.ac.jp     2.8.16     Japan Adv. Institute of Sci & Tech
  hemp.imel.kyoto-u.ac.jp 2.8.16     Kyoto University
  scorpio.cse.kyutech.ac.jp 2.8.16   Kyushu Institute of Tech., IIZUKA
  murasame.ics.es.osaka-u.ac.jp 2.7.2g Osaka
  hohoi.cs.titech.ac.jp   2.8.16     Tokyo Institute of Technology
  akiu.gw.tohoku.ac.jp    2.8.16     Tohoku University
  utsun.s.u-tokyo.ac.jp   2.7.2g     University of Tokyo
  irc.ube-c.ac.jp         2.8.16     Dept of Computer Science, Ube College
  toumon.cfi.waseda.ac.jp 2.8.16     Waseda University, Tokyo
  endo.wide.ad.jp         2.8.16     WIDE-NOC of Fujisawa
  hamlet.wg.omron.co.jp   2.8.16     OMRON Corporation, Kyoto
  omrongw.wg.omron.co.jp  2.8.16     OMRON Corporation, Kyoto
  scslwide.sony.co.jp     2.7.2g     Sony Computer Science Laboratory, Inc.
  biscuit.mmws.astem.or.jp 2.7.2g    ASTEM RI, Kyoto
  choshi.kaba.or.jp       2.7.2g     KABA, Kyoto
  wings.tokai-ic.or.jp    2.8.16     TRENDY IRC

Korea

  cbubbs.chungbuk.ac.kr   2.7.2g     Chung-Buk Univ. BBS
  ns.kaist.ac.kr          2.7.2g     - KAIST
  han.hana.nm.kr          2.7.2g     HANA NET IRC SERVER

-----< Australia >----------------------------------------------------------

  aquarius.cssc-woll.tansu.com.au 2.8.16 Telecom Australia Server
  yoyo.cc.monash.edu.au   2.8.16     Monash University [The Biggest Uni In OZ!]
  yamabico.cs.uow.edu.au  2.8.16     Uni. of Wollongong
  rabble.uow.edu.au       2.8.14     Uni. of Wollongong
  irchat.utas.edu.au      2.8.10     University of Tasmania
  tartarus.ccsd.uts.EDU.AU 2.7.2g.troy6+wallops UTS (Sydney Aust) [I'm an SGI and I'm okay]
  notjules.itd.uts.edu.au 2.8.16     Finally a 2.8* version that works ?
  fox-in.socs.uts.edu.au  2.8.14     As if sbug didnt have enough memory problems...
  gwen.st.nepean.uws.edu.au 2.8.14   An AT Standalone
  kay.st.nepean.uws.edu.au 2.8.14    a machine which struts
  defiance.vut.edu.au     2.8.16     Victoria University OF Technology
  ircserver.cltr.uq.oz.au 2.8.15+hh  HulkHoganised Server, Queensland
  jello.qabc.uq.oz.au     2.8.16+hh  University of Queensland

-----< Europe >-------------------------------------------------------------

Austria

  uni-linz.ac.at          2.8.12     Johannes Kepler University Linz
  itc.univie.ac.at        2.8.16     Alma mater Rudolphina
  olymp.wu-wien.ac.at     2.8.16     Vienna

Belgium

  OtHelLo.UlB.aC.Be       2.8.16.dl1 Ecole Polytechnique (ULB-Bruxelles)
  is1.bfu.vub.ac.be       2.8.16.    Vrije Universiteit Brussel

Croatia

  smile.srce.hr           2.8.16     University Computing Centre Zagreb

Czech Republic

  irc.praha.cz            2.8.16     Praga Caput Regni, Bohemia

Denmark

  alk.iesd.auc.dk         2.8.16     Aalborg University, IESD

Finland

  irc.eunet.fi            2.8.16     EUnet
  irc.funet.fi            2.8.17pre8+V2 Finnish University and Research Network (FUNET)
  hutirc.cs.hut.fi        2.8.14+op  Helsinki University of Technology, CS Lab
  irc.cs.hut.fi           2.8.16     Helsinki University of Technology, CS Lab
  freenet.hut.fi          2.8.16     Freenet
  cs.jyu.fi               2.8.16     University of Jyvaskyla
  aapo.it.lut.fi          2.8.16     Lappeenranta University of Technology
  gopher.it.lut.fi        2.8.17pre10+3 Lappeenranta University of Technology
  kannel.lut.fi           2.8.16     Lappeenranta University of Technology
  irc.otol.fi             2.8.16     Oulu Institute of Technology
  rieska.oulu.fi          2.8.16     Unixverstas Olutensin, Finlandia Vodka
  tolsun.oulu.fi          2.8.16     Unixverstas Olutensin, Finlandia Vodka
  irc.ttl.fi              2.8.16     Turku Telephone Company
  irc.cc.tut.fi           2.8.14     TUT Computer Centre
  vehka.cs.uta.fi         2.8.15     University of Tampere
  polaris.utu.fi          2.8.14     University of Turku
  brando.uwasa.fi         2.8.16     University of Vaasa

France

  Eurecom8.Cica.FR        2.8.16+chef Institut Eurecom
  cnam.cnam.fr            2.8.16     Conservatoire National des Arts et Metiers
  Irc.ENST.Fr             2.8.17pre10 Irc.ENST.Fr [137.194.160.46], Paris
  irc.obspm.fr            2.8.16.dl1-UDP_SUX Observatoire de Paris, Meudon
  sil.polytechnique.fr    2.8.16+Chef Ecole Polytechnique, Palaiseau
  Firmin.Labri.U-Bordeaux.Fr 2.8.16+Chef+Ircops_are_lame_froggies_uber_alles_fuck_the_finns Universite Bordeaux I
  Dafne.Mines.U-Nancy.FR  2.8.16     Ecole des Mines de Nancy
  CISMhp.Univ-Lyon1.FR    2.8.16.N#1 CISM - INSA de Lyon
  PropeServer.Univ-Lyon1.FR 2.7.2g.Nono+4 INSA Lyon - "Prope" Server

Germany

  noc.belwue.de           2.8.14w    BelWue
  fu-berlin.de            2.8.14     Freie Universitaet Berlin
  TU-Muenchen.DE          2.7.2g     TU Muenchen
  Uni-Erlangen.DE         2.8.16     The 'Berch'-Server
  uni-karlsruhe.de        2.8.16.cf2 University of Karlsruhe
  Uni-KL.DE               2.8.16     Kaiserslautern University
  Uni-Paderborn.DE        2.8.16     University of Paderborn
  Uni-Rostock.DE          2.8.16     Rostock
  uni-stuttgart.de        2.8.16     Stuttgart University
  rwth-aachen.xde         2.8.14yegg/argv Aachen

Hungary

  darmol.elte.hu          2.8.16     Eotvos University of Budapest

Iceland

  isgate.is               2.8.14+op  ISnet

Italy

  ghost.dsi.unimi.it      2.8.16.dl1 University of Milan
  sun02.ccii.unipi.it     2.8.16     University of Pisa

The Netherlands

  ircserver.et.tudelft.nl 2.8.16.dl1 Delft University of Technology
  svbs01.bs.win.tue.nl    2.8.16     Technical University of Eindhoven

Norway

  alf.uib.no              2.8.16     Universitas Bergensis, Ordinateurus Centre
  mimir.ifi.uio.no        2.8.16     Universitas Osloensis
  irc.cs.uit.no           2.8.16     University of Tromsoe
  irc.nvg.unit.no         2.8.17pre8 Nettverksgruppa - Universitas Nidrosiensis

Poland

  galaxy.uci.agh.edu.pl   2.8.16     AGH Krakow
  irc.fuw.edu.pl          2.8.16     Warsaw University, Warsaw

Russia

  irc.demos.su            2.8.16dog3 DEMOS/* Russian IRC-Server

Slovenia

  irc.arnes.si            2.8.16     Fastest growing userbase in Europe!

Sweden

  gwaihir.dd.chalmers.se  2.8.16     Chalmers Tekniska Lekskola, G|teborg
  Saturnus.pt.hk-r.se     2.8.16.Ace SoftCenter, Ronneby
  irc.nada.kth.se         2.8.16     Royal Institute of Technology, Stockholm
  krynn.efd.lth.se        2.8.12     Lund Institute of Technology

Switzerland

  disuns2.epfl.ch         2.8.16     Swiss Fed Inst of Tech of Lausanne
  firewall.epfl.ch        2.7.2g     Swiss Fed. Inst. of tech of Lausanne
  irc.ethz.ch             2.8.16     Swiss Fed Inst of Tech of Zurich
  ircserv.imp.ch          2.8.12     Improware AG, Fuellinsdorf
  cuisund.unige.ch        2.8.16.NOTE University of Geneva

United Kingdom

  serv.eng.abdn.ac.uk     2.8.16     University of Aberdeen
  shrug.dur.ac.uk         2.8.16     University of Durham
  stork.doc.ic.ac.uk      2.8.15     Dept of Computing, Imperial College, London
  fennel.compnews.co.uk   2.8.9      PressNet Hub
  dismayl.demon.co.uk     2.8.16     Demon Internet, London

-----< North America (except US) >-----------------------------------------

Canada

  ug.cs.dal.ca            2.8.14.moose Official CoD Sponsor. Halifax, Nova Scotia
  fox.nstn.ns.ca          2.7.2g     NSTN Inc, Halifax, Nova Scotia
  von-neumann.info.polymtl.ca 2.8.14.cp.w Montreal Polyserver
  castor.cc.umanitoba.ca  2.8.16.Wlps.Sln.Max200 University of Manitoba
  degaulle.hil.unb.ca     2.8.16.wallops University New Brunswick, Fredericton, NB
  clique.cdf.utoronto.ca  2.8.16.Wlp.Sln.Wzd University of Toronto
  irc.yorku.ca            2.8.16.WALLOPS York University's IRC Server
  io.org                  2.8.16     Internex Online

Mexico

  next00.mty.itesm.mx     2.8.12     ITESM Campus Monterrey, Mexico Hub
  irc.pue.udlap.mx        2.8.16     Universidad de Las Americas, Puebla

-----< South America >-----------------------------------------------------

Chile

  UNIRED                  2.8.16.U2  El primer irc-server de SUDAMERICA

-----< United States >-----------------------------------------------------

Alaska

  merlin.acf-lab.alaska.edu 2.8.16.mal-2 The Small Talk Stinks Alaskan Server

Arkansas

  irc.engr.uark.edu       2.8.16     Home of the Razorbacks

California

  irc.netsys.com          2815p2.rw+cp NETSYS COMMUNICATIONS
  nova.unix.portal.com    2.8.15pre2.rw Cupertino
  harp.aix.calpoly.edu    2.8.16     The SLOer than you server...
  irc.caltech.edu         2.8.16     California Institute of Technology
  B-w6yx.stanford.edu     2.8.16     From DC to Daylight
  w6yx.stanford.edu       2.8.16     From DC to Daylight
  hyperion.cs.ucdavis.edu 2.8.16dog3 University of California at Davis
  othello.ucdavis.edu     2.8.14cp   University of California, Davis
  ucsd.edu                2.8.16     Univ of Calif San Diego

Colorado

  irc.Colorado.EDU        2.8.16     Univ of Colorado Server

Florida

  excalibur.mlb.semi.harris.com 2.8.16 Warlord Server, Melbourne
  pegasus.cc.ucf.edu      2.8.16     University of Central Florida
  infrared.creol.ucf.edu  2.8.10+sw  University of Central Florida
  irc.math.ufl.edu        2.8.16     The Happiest Server on IRC

Illinois

  irc.acns.nwu.edu        2.8.16     Northwestern University, Evanston
  necromancer.ece.uiuc.edu 2.8.16    UIUC Client Server
  irc.uiuc.edu            2.8.16     University of Illinois at Urbana-Champaign

Indiana

  Irc.Indiana.Edu         2.8.16.atdp Indiana University Primary IRC Server

Iowa

  ircserver.iastate.edu   2.8.16.zephyr.atdp Iowa State Alternate Reality Server

Kentucky

  irc-host.ms.uky.edu     2.8.16     The Hole in the Rock Server at the Univ. of KY

Maryland

  irc.digex.net           2.8.16     Digital Express Group

Massachusetts

  world.std.com           2.8.16     The World @ Software Tool & Die
  berry.cs.brandeis.edu   2.8.10     Brandeis University IRC Server
  csa.bu.edu              2.8.15     Boston University, Boston
  husc10.harvard.edu      2.8.16     If we're up, then all is well ;>
  irc.mit.edu             2.7.2g     MIT Project Athena
  irc-2.mit.edu           2.8.16dog3 Mass. Institute of Technology, Cambridge
  wpi.wpi.edu             2.7.2g     Worcester Polytechnic Institute

Michigan

  hobbes.kzoo.edu         2.8.16     Kalamazoo College
  pegasus.ccs.itd.umich.edu 2.8.14+cp-dog3 An Inbred-Jed Production
  coyote.cs.wmich.edu     2.8.14     WMU,   Kalamazoo

Minnesota

  irc.tc.umn.edu          2.8.16     University of Minnesota, Twin Cities

Missouri

  sluaxa.slu.edu          2.8.16     St. Louis University

New Jersey

  hertz.njit.edu          2.8.16     NJIT Computing Services Department,Newark
  pilot.njin.net          2.8.16.pilot Rutgers Univeristy New Brunswick

New Mexico

  IRCServer.SantaFe.EDU   2.8.16     Santa Fe Institute

New York

  azure.acsu.buffalo.edu  2.8.16+2   University at Buffalo, Buffalo
  red-dwarf.cit.cornell.edu 2.8.16   The CIT IRC
  mcphy0.med.nyu.edu      2.8.16     The Mid-Manhattan Link
  colossus.cs.rpi.edu     2.8.16     Rensselaer Polytechnic Institute IRC Server
  netserv2.its.rpi.edu    2.8.16     Rensselaer Polytechnic Institute, Troy
  irc.rutgers.edu         2.8.16     NONE
  mothra.syr.edu          2.8.16     Syracuse University

North Carolina

  hobbes.catt.ncsu.edu    2.8.14     North Carolina IRC Server

Ohio

  slc9.ins.cwru.edu       2.7.2i.1   CWRU IRC Server

Oklahoma

  lincoln.ecn.uoknor.edu  2.8.16     SOONER HUB SERVER 2.8.16,  SERVING SOONERS

Oregon

  irc.csos.orst.edu       2.8.16     The Beaver Server of The Northwest

Pennsylvania

  irc.duq.edu             2.8.16     Duquesne University, Pittsburgh
  chestnut.chem.upenn.edu 2.8.16     University of Pennsylvania QUANYIN Server

Texas

  ra.oc.com               2.8.16     OpenConnect Systems Server, Dallas
  acme.etsu.edu           2.7.2g.ID.herm The Magical Realms Server
  irc.tamu.edu            2.8.16     Texas A&M University IRC server
  omega.tamu.edu          2.8.16     Da' Maroon Lagoon of Aggieland!
  dewey.cc.utexas.edu     2.8.16     UT's "I Repeat Class" server
  lonestar.utsa.edu       2.8.16     LoNeStAr Server at Univ of TEXAS at SA [UTSA]

Utah

  hamblin.math.byu.edu    2.8.11     Brigham Young University Math Dept

Virginia

  beta.cs.jmu.edu         2.8.16     James Madison University CS
  muselab.ac.runet.edu    2.8.16     Radford University
  poe.acc.Virginia.EDU    2.8.16     University of Virginia

Washington

  irc.eskimo.com          2.8.16jayk Eskimo North. Located in Seattle
  goren1.u.washington.edu 2.8.14.wall.ban.cp.users.(summon).slug.slime.rust We're better than you, so nyah.
  goren2.u.washington.edu 2.8.16.users.ban.wallops.dawg.food.woof We're better than you, so nyah.

Washington, D.C.

  eff.org                 2.8.16     Electronic Frontier Foundation, G. St NW

----< Version Information >-------------------------------------------------

Total    By Version      Count     Version

                           13      2.7.2g
                            1      2.7.2g.ID.herm
                            1      2.7.2g.Nono+4
                            1      2.7.2g.troy6+wallops
             17             1      2.7.2i.1

              1             1      2.8.9

                            2      2.8.10
              3             1      2.8.10+sw

              1             1      2.8.11

              4             4      2.8.12

                            9      2.8.14
                            1      2.8.14+cp-dog3
                            2      2.8.14+op
                            1      2.8.14.cp.w
                            1      2.8.14.moose
                            1      2.8.14.wall.ban.cp.users.(summon).slug.slime.rust
                            1      2.8.14cp
                            1      2.8.14w
             18             1      2.8.14yegg/argv

                            3      2.8.15
                            1      2.8.15+hh
                            1      2.8.15pre2.rw
              6             1      2815p2.rw+cp

                          104      2.8.16
                            1      2.8.16+2
                            1      2.8.16+Chef
                            1      2.8.16+Chef+Ircops_are_lame_froggies_uber_alles_fuck_the_finns
                            1      2.8.16+chef
                            1      2.8.16+hh
                            1      2.8.16.
                            1      2.8.16.Ace
                            1      2.8.16.N#1
                            1      2.8.16.NOTE
                            1      2.8.16.U2
                            1      2.8.16.WALLOPS
                            1      2.8.16.Wlp.Sln.Wzd
                            1      2.8.16.Wlps.Sln.Max200
                            1      2.8.16.atdp
                            1      2.8.16.cf2
                            3      2.8.16.dl1
                            1      2.8.16.dl1-UDP_SUX
                            1      2.8.16.mal-2
                            1      2.8.16.pilot
                            1      2.8.16.users.ban.wallops.dawg.food.woof
                            1      2.8.16.wallops
                            1      2.8.16.zephyr.atdp
                            3      2.8.16dog3
            132             1      2.8.16jayk

                            1      2.8.17pre10
                            1      2.8.17pre10+3
                            1      2.8.17pre8
 186          4             1      2.8.17pre8+V2

----< Other Servers List >--------------------------------------------------
 
These are servers which are not for general client use, or which have
'test' or 'experimental' in their title; these active servers are not
counted in the Version List above.

sifon.cc.mcgill.ca                  McGill University, Computing Centre IRC Server, 2.
irc.services.ca                     Canadian IRC Services Server
garnet.msen.com                     Msen IRC Test Server
jobe.shell.portal.com               Portal 2.8.x Test Server
service.de                          TU Muenchen
polaris.ctr.columbia.edu            Columbia University Testing
Irc-1.Indiana.Edu                   Indiana Experiemental IRC Server
note.mit.edu                        MIT NOTE server
service.rutgers.edu                 RUTGERS University Note Service
test.mines.u-nancy.fr               Ecole des Mines de Nancy, France
service.obspm.fr                    DL's Test Service
test.biscuit.mmws.astem.or.jp       Kyoto
cent1.lancs.ac.uk                   NONE
monitor.us                          The US IRC Monitor

----< Delete List >---------------------------------------------------------
 
These currently inactive servers will be deleted from the next list unless
they appear on the IRCnet again.

cairo.anu.edu.av                    Coombs IRCD (IRC's adopted home) ANU
polytech.ulb.ac.be                  Ecole Polytechnique (ULB-Bruxelles)
jupiter.sun.csd.unb.ca              University New Brunswick, Fredericton, NB
eskinews.eskimo.com                 Eskimo North-Seattle
irc.lm.com                          Telerama IRC Server
irc.dknet.dk                        DKnet IRC Server
athens.caltech.edu                  California Institute of Technology
gsusgi2.gsu.edu                     Georgia State University, Atlanta the Olympic City
IRC-2.INDIANA.EDU                   Nickel.ucs.indiana.edu[129.79.10.5] 2.8.16+patches
hub.cs.jmu.edu                      James Madison University CS Hub
freedom.nmsu.edu                    New Mexico State U. NMSU SCF
old272.stanford.edu                 Stanford 2.7.2 Old Server
irc.stolaf.edu                      St. Olaf College, Northfield
mole.cis.ufl.edu                    Simulation Server
llull.math.ufl.edu                  The Happiest Server on IRC
lehtori.cc.tut.fi                   Tampere University of Technology
HpLyot.ObsPM.Circe.FR               Observatoire de Paris, Meudon
Hector.Labri.U-Bordeaux.Fr          Universite Bordeaux I
samos.aegean.ariadne-t.gr           University of the Aegean
areti.naval.ntua.gr                 N.A.M.E.  N.T.U.  ATHENS
cyprus.csd.uch.gr                   The Greek Server!
irc.datasrv.co.il                   DataServe LTD.
cdc835.cdc.polimi.it                Politecnico di Milano
sun.cisi.unige.it                   Universita' di Genova
sun02.iet.unipi.it                  University of Pisa, Italy
scpower2.sfc.keio.ac.jp             KEIO SFC IRC SERVER
totto.ics.kula.kyoto-u.ac.jp        Faculty of Integrated Human Studies, Kyoto Univers
siratuyu.ics.es.osaka-u.ac.jp       Osaka
hamlet.nff.ncl.omron.co.jp          OMRON Corporation, Kyoto, Japan
noc.uci.agh.edu.pl                  AGH Krakow, Poland
irc.tuzvo.sk                        TU Zvolen
irc.nchu.edu.tw                     Computer Center at NCHU, Taiwan, ROC
irc.csie.nctu.edu.tw                IRC of CSIE, NCTU, Taiwan, ROC
im.mgt.ncu.edu.tw                   IM/MGT at NCU, Taiwan, ROC
IRC.NSYSU.edu.tw                    IRC of National Sun Yat-sen University
NEWS.NSYSU.edu.tw                   IRC of National Sun Yat-sen University
ccms.ntu.edu.tw                     National Taiwan University


-- 
 |    _   _  _|    \  /  _   _|  _  _   #  "... Podran cortame las manos
 |__ (_) |  (_|     \/  (_| (_| (= | 	#   pero nunca mi pensamiento .."
                                        #  " Cachurero de la Internet ..."
 <epedegu@inf.utfsm.cl>, U. Santa Maria	#  Gabriel Epeldegui T. 

-----------[000310][next][prev][last][first]----------------------------------------------------
Date:      18 May 1994 21:07:18 GMT
From:      smiles@powerdog.com (Kevin Ruddy)
To:        comp.dcom.lans.ethernet,comp.protocols.tcp-ip
Subject:   Bridging software

I'm sorry about the newsgroups; these were the only two I could think of
that might have the answer.

I'm looking for public domain, freely-available, or commercial bridging
software -- but is must contain source.  I've heard of KarlBridge and I know
there are others, but I can't find any references around anymore.  I've
tried looking at rtfm.mit.edu to no avail.

Thanks for any assistance.
-- 
kevin.ruddy@powerdog.com

-----------[000311][next][prev][last][first]----------------------------------------------------
Date:      19 May 94 01:38:06 GMT
From:      rtkao@remus.rutgers.edu (Richard Kao)
To:        comp.protocols.tcp-ip,comp.protocols.tcp-ip.ibmpc
Subject:   KA9Q with dial-in slip.


I have been trying to setup a dialin slip server and I have it working
after some work.  However, I had to use a second IP address in our
existing network which I didn't register with our network
administrator.  I would like to setup the slip server as follows...

 ------------------                            -----------------
| Existing Network |      ---------------     | New Domain for  |
|                  |-----| KA9Q SLIP     |----| SLIP Connetions |
| IP 141.202...    |     | Server/Router |    | IP 141.200...   |
 ------------------       ---------------      -----------------

I wan't the SLIP connections to have a completly different address.
This way I don't have to register it with our network people.  Is it
possible to do this with out changing anything in our existing
network.  Is it possible to telnet to IP address 141.202.200.1 from a
slip connection IP address 141.200.200.1.  Does anyone have any ideas?

-----------[000312][next][prev][last][first]----------------------------------------------------
Date:      Thu, 19 May 1994 03:28:54 GMT
From:      fitz@wang.com (Tom Fitzgerald)
To:        comp.protocols.tcp-ip
Subject:   Re: Duplicate IP address detection

ericd@gaudi.csufresno.edu (Eric Douglas) writes:

> Re: Finding duplicate IP addresses on LAN. If your machines
> which possibly have the same IP support a full TCP stack, 
> it seems possible that opening a TCP connection to an arbritrary
> port (there must be a server listening on it) then checking the
> number of SYN/ACK replies would give you a fairly accurate number.

No, because the SYN isn't broadcast.  The originating system will broadcast
an ARP, get back multiple responses from all the systems with the same IP
address, pick one (usually the last to respond), and unicast the SYN packet
to it.  You'll never get more than one SYN/ACK.  There isn't any easy way
of counting the number of ARP responses.

Watching ARP tables is usually the best way of finding duplicates and
people using unassigned IP addresses.

-- 
Tom Fitzgerald   Wang Labs   Lowell MA, USA   1-508-967-5278   fitz@wang.com
Pardon me, I'm lost, can you direct me to the information superhighway?

-----------[000313][next][prev][last][first]----------------------------------------------------
Date:      19 May 1994 04:59:46 GMT
From:      lirwin@silicon.csci.csusb.edu (Loren Irwin)
To:        comp.protocols.tcp-ip
Subject:   SLIP Connection and Routing question



I have a SLIP connection into a terminal server at school,
using FreeBSD on my home machine.  The problem is that
I'm isolated to the machines at school; If I want to telnet
or ftp to a remote site, I have to login to one of the machines
on campus first.  So I can't run things like mosaic directly on
my machine.  I can't get mail directly here either.

Apparently, the routes are setup this way to prevent people from 
accessing remote sites directly from the terminal server prompt.

Is there a way to bypass the routing on the terminal server
and direct packets to one of the other routers on campus? I can
ping the other router.  Is it just a matter of doing something
with routed?


Thanks a lot.


-------
Loren Irwin  	lirwin@silicon.csci.csusb.edu


-----------[000314][next][prev][last][first]----------------------------------------------------
Date:      19 May 1994 17:30:11 -0700
From:      lstowell@pyrnova.mis.pyramid.com (Lon Stowell)
To:        comp.protocols.tcp-ip
Subject:   Re: CTRL-ALT-DEL problem with sockets betn. PC and Unix

In article <sivapras.769378914@shazam.cs.iastate.edu> sivapras@cs.iastate.edu (Gowri Sankar Sivaprasad) writes:
>
>Hello,
>
>	I am facing a problem with UNIX sockets.
>	
>	Problem:  I establish a TCP connection between my SCO UNIX system and a
>	Windows system (486 dx).  Then I reboot the windows m/c (CTRL_ALT_DEL).
>	But the SCO system does not recognize that the client has gone down. 
>	If I do a netstat on the SCO system, I find the socket to that client as
>	"ESTABLISHED".  This is a major problem because it eats up TCP ports.
    It is considered proper behavior on a socket to CLOSE the socket
    before you reboot the machine(s) on the ends of the socket.
    Unix folks tend to do this more, as the Unix type file systems
    tend not to survive unplanned power downs and resets very well.

    Unfortunately there are now socket partners who don't send RST
    on a socket before dying.  The protocol does NOT require that
    anything go back and forth on a socket, so how on earth would the
    partner know that one partner is in trouble or has a "somewhat
    common sense challenged" operator?  
>	
>	Please suggest some solutions.

    Whether you should use keepalives or application level polling
    on the socket to detect a dead connection is almost a religious
    issue.  The best answer will depend on how much of the programming
    environment is under your control and how good (and tuneable)
    the keepalive implementation on your TCP stacks is.  
>
>	PLEASE E-MAIL ME SINCE I DON'T READ THIS GROUP OFTEN.

    This issue is discussed regularly on this group, perhaps you may
    want to browse more frequently?  You would already be aware of the
    issue. 

    Just look for subjects with "Keepalives--were they really meant
    for DOS PC's and their braindead operating systems and
    intellectually challenged operators" or just the word "keepalive" 
    in the header.



-----------[000315][next][prev][last][first]----------------------------------------------------
Date:      Thu, 19 May 1994 06:36:22 GMT
From:      mark@cyantic.com (Mark T. Dornfeld)
To:        comp.protocols.tcp-ip
Subject:   Using RARP for IP number assignment

One of our clients would like to centralize the administration of their IP
network, and would like to start by having IP numbers assigned by a central
server.  The systems that require assignment are primarily Macs running
MAC/TCP and have the capability of getting their IP numbers assigned from a
server using RARP (I assume).  The server will probably be an SCO ODT
machine and I don't yet have access to it.  I have some questions regarding
this procedure.

I know that the file /etc/ethers must be set up and maintained on the
server, but is there a process that must be started in order for the server
to respond to a RARP request? 

Does SCO ODT have the capability to serve IP numbers?

Are there any pitfalls in going this route that I should warn them of now
before they jump in?

Does DHCP address this issue completely?

Will running a DNS server make this any easier?

Thanks in advance for your help.
-- 

Mark T. Dornfeld, Cyantic Systems Corporation       Voice: (416) 621-6166
1 Eva Road Suite 301                            Facsimile: (416) 621-6212
Etobicoke, Ontario, M9C 4Z5 CANADA                  Email: mark@cyantic.com

-----------[000316][next][prev][last][first]----------------------------------------------------
Date:      19 May 1994 07:09:48 GMT
From:      Anders Liljegren <Anders.Liljegren@udac.uu.se>
To:        comp.protocols.appletalk,comp.protocols.tcp-ip,comp.sys.mac.comm,fr.network.divers
Subject:   Re: MacTCP 2.0.4 & BOOTP RFC 1048 & DNS

In article <1994May18.161546.5921@genethon.fr> Claude Scarpelli,
claude@genethon.genethon.fr writes:

> I try to configure MacTCP 2.0.4 with BOOTP. BootP server is running on
> SunOS4.1.3, and MacTCP is running on a Quadra 650 (System 7.1 french).
> 
...
 
> 
> The Mac record its IP address, the gateway address and the subnet mask
> correctly. But it does not get neither the domain name nor the name
> server adress.

I'n not shure about version 2.0.4 of MacTCP, but by experimenting 
I've arrived att the following conclusions for earlier versions:

It will not get the domain name address. There is really no reason.
It should get this from the domain name server. In the bootp table
it's only used as an identifier.

It will however get _one_ domain name server. If you give several in
the bootp table it will only use one (the last one if I remember 
correctly). Also, it will not enter the domain name server into 
the MacTCP control panel, but it _will_ use it. Having only one 
domain name server configured is of course not very nice. Not much 
will work when it's down.

There are also some other stupid bugs in MacTCP. E.g., it will only
get one MX record from a domain name server (the last one ?).

The user interface for entering domain name servers manually is not
very good (I'm working hard here to refrain from using bad 
language). If you want to configure MacTCP for using two domain name
servers, one of which is used normally and the other is used if 
the first is down, you should do it in the following way. Don't
ask me why! It's to complicated to explain :-)

>  default.domain    xxx.xxx.xxx.xxx     x
>  .                 xxx.xxx.xxx.xxx     o
>  .                 yyy.yyy.yyy.yyy     o

where "deafult.domain" is the domain you mostly connect to, 
"xxx.xxx.xxx.xxx" is the IP address for the first hand choice for
domain name server, and "yyy.yyy.yyy.yyy" is the IP address for
the other domain name server. (The angle brackets should NOT be
entered and are just there to avoid that some stupid mailers 
breaks the letter at the periods.)

All this (and more) has been reported to Apple many times. Why they 
haven't fixed it long since is beyond me.

-------------------------------------------------------------------
Anders Liljegren, universitetsenheten       Mail: UDAC
Phone:  +46 18 18 77 51                           Box 174
Beeper: +46 746 47 52 09                          S-751 04 UPPSALA
Fax:    +46 18 51 66 00                           Sweden
E-mail: Anders.Liljegren@UDAC.UU.SE
-------------------------------------------------------------------

-----------[000317][next][prev][last][first]----------------------------------------------------
Date:      19 May 1994 08:27:12 GMT
From:      bortz@cnam.cnam.fr (Stephane Bortzmeyer)
To:        comp.protocols.tcp-ip
Subject:   Re: Duplicate IP address detection

In article <1994May17.172103.5634@ivax>, harvey@indyvax.iupui.edu (James Harvey) writes:
...
>This can be a continual problem on a large net.  For example here at IUPUI
>recently we had at least 105 duplicates out of 4,251 addresses in use.
>You can obtain this kind of information by periodically polling the ARP tables
>on your routers.  

A quick and dirty Perl script to do it (on every router/host which 
supports SNMP) is:

ftp://ftp.cnam.fr/pub/CNAM/MISC/record_arp.tar.Z

(It needs the Tricklet package, see the README at the end of this message.)

                    
            Record addresses in a LAN from SNMP agents

This set of Perl scripts allow you to record addresses of machines
which "speak" on your LAN.

It does so by dumping ARP tables of SNMP agents. ARP tables are tables of 
mapping between a MAC (Medium Access Control, hardware) addrress, such as an 
Ethernet address and a network (high-level) address, at the present time, 
only an IP address.

To use it, you need:

- Perl, a nice programming language for system and network 
administrators. Perl is available on uunet.uu.net:gnu/perl-xxx.tar.Z, on
tut.cis.ohio-state.edu:perl/xx and on jpl-devvax.jpl.nasa.gov:pub/perlxxx.
It is described in "Programming Perl" by Wall and Schwartz (O'Reilly).

- at least one SNMP agent on a frequented machine (I mean a machine
which speaks with many others, such as a central router). This 
agent has to be in MIB-II (RFC 1213). Today, almost all network devices
(hubs, routers, bridges, etc) are sold with a built-in SNMP agent. It can
also run on a general-purpose computer too. SNMP is described in "The simple
book" by Rose (Prentice-Hall).

- a software package which allows you to direct SNMP requests from
the command line. My scripts are tailored for Tricklet, but it 
shoudln't be too hard to modify them for, say, CMU's SNMP.

To get Tricklet, a great package to do network management with SNMP, ftp to 
dutepp0.et.tudelft.nl:/pub/Tricklet. There are many other interesting network
things on this machine, all made by the Data Network Performance Analysis 
Project at Delft University of Technology. Many thanks to them.
For CMU, try lancaster.andrew.cmu.edu:pub/cmu-snmpxxx.tar. 

You have the following scripts:

- arp.pl dumps the ARP table from a distant machine. Useful to test if this
machine has a correct SNMP agent.

- record_mac_addresses.pl is the most important script: it gathers information
from your agents. Edit the line beginning with "@agents=" to write your own
agents. Choose a name for your database and edit "$DBM_file_name =" and 
"$log_file_name =" accordingly. Change the "$mail" command to put the appropriate
address. Then, run this script from crontab every three or four hours.

- print_mac_addresses.pl dumps the whole database. I like the following C 
shell alias "alias macaddr 'perl /path/print_mac_addresses.pl | grep -i \!$'"
which allows me to find a machine by its name, IP address or MAC address.

- remove_mac_address.pl to suppress an entry (by its IP address).

Of course, this package is without any warranty and I don't
speak for my employer...

(But if you have bug reports, I'd like to see them.)

Stephane Bortzmeyer           Conservatoire National des Arts et Metiers	
bortzmeyer@cnam.cnam.fr       Laboratoire d'Informatique
                              292, rue Saint-Martin			
tel: +33 (1) 40 27 27 31      75141 Paris Cedex 03
fax: +33 (1) 40 27 27 72      France	

"C'est la nuit qu'il est beau de croire a la lumiere." E. Rostand



-----------[000318][next][prev][last][first]----------------------------------------------------
Date:      Thu, 19 May 94 16:37:23 PDT
From:      sjolander@minerva.com
To:        comp.protocols.tcp-ip
Subject:   TFTP Server for the Macintosh


Can anyone tell me where I can obtain a TFTP Server
for the Macintosh?
Thanks,
Stephen Sjolander
sjolander@minerva.com


-----------[000319][next][prev][last][first]----------------------------------------------------
Date:      Thu, 19 May 94 19:17:03 PDT
From:      cclarke%ghs.uucp@usc.edu (Chris Clarke)
To:        comp.protocols.tcp-ip
Subject:   Is there a PD version of TCP/IP & NFS for UNIX out there?

  To: comp.protocols.nfs
      comp.protocols.tcp-ip
      comp.protocols.tcp-ip.ibmpc
Date: 5-19-94

                     Looking for TCP/IP/NFS/PPP for UNIX
                     -----------------------------------

Does anyone know if there any public domain implementations of TCP/IP/NFS/PPP 
for UNIX out there?

Have any been ported to SCO UNIX?

All the best.

Chris Clarke
cclarke%ghs.uucp@usc.edu

-----------[000320][next][prev][last][first]----------------------------------------------------
Date:      19 May 1994 19:12:21 -0400
From:      rdavis4@umbc.edu (davis robert)
To:        comp.protocols.tcp-ip
Subject:   Question about TCP/IP code for binary file x-fer

While trying to get some old TCP/IP code working for an old computer
system, I've come across a slight problem: it is only for ASCII files.
Could someone tell me, basically, what types of changes/additions I
need to make to add binary file transfer capability to it?  Thanks in
advance for any information that anyone can provide about this!

-- 
Robert D. Davis      |          Eccentrics have more fun! :-) 
...uunet!mystica!rdd |------------------------------------------------------
rdavis4@umbc.edu     | Looking for work... if you can hire me, please ftp my
1-410-744-7964       | resume from ftp.digex.net, /pub/access/rdd/resume.

-----------[000321][next][prev][last][first]----------------------------------------------------
Date:      Thu, 19 May 1994 13:13:52 GMT
From:      mark@cyantic.com (Mark T. Dornfeld)
To:        comp.protocols.tcp-ip
Subject:   Assigning IP addresses to a CISCO WAN

Below is a diagram of LAN/WAN setup. There are two local area networks
connected by two CISCO IGS Routers via 56K synchronous lines. We currently
are having a discussion about how to assign IP numbers to the WAN
Synch ports.

Each LAN has its own class C IP subnet, and currently the two synchronous
ports comprise a third class C subnet. 

LAN1     ---------------------------------     192.246.150.x
                       |
                    --------
                    |  E1  |
           CISCO1   |  S1  |    192.246.151.1
                    --------
                       |
      56K SYNC WAN     |
                       |
                    --------
                    |  S2  |    192.246.151.2
           CISCO2   |  E2  |
                    --------
                       |
LAN2     ---------------------------------     192.246.152.x

Is this the correct way to number the WAN?  Should the WAN have it's own
subnet, or should interface S1 have an address from LAN2 and interface S2
have a number from LAN1.  It seems that both ways work in different
situations with other vendors hardware (Telebit), but are there problems
with either approach that are not immediately evident?

What is the preferred approach?

Ultimately we would like to use a subnetted Class C number for the WAN in
order not to waste 254 IP numbers on a two-port WAN if that is the right
way to set it up.

Please respond by email, or post if you think this is worth discussing.

Thanks
-- 

Mark T. Dornfeld, Cyantic Systems Corporation       Voice: (416) 621-6166
1 Eva Road Suite 301                            Facsimile: (416) 621-6212
Etobicoke, Ontario, M9C 4Z5 CANADA                  Email: mark@cyantic.com

-----------[000322][next][prev][last][first]----------------------------------------------------
Date:      Thu, 19 May 1994 13:51:49 GMT
From:      bob@astph (Bob Ford)
To:        comp.protocols.tcp-ip
Subject:   Measuring Network Throughput

We are scouting about for freeware/shareware that will help us to
measure the data throughput of our client/server application.  We have
completed the functionality of our application and now have hopes of
improving the speed of the thing.  We can easily measure speeds of the
client and server processes using "time" and "prof".  However, we would
also like a network monitoring utility that gives us a bytes/min
measure of the data flowing across the network from our application.

We already have quite a number of free network tools such as ttcp,
sock, and Netwatch.  While these are all quite useful, none of them
does quite what we want.  We would appreciate any pointers to software
that would help us make these data flow measurements.

We are running ISC UNIX version 4.0  on 386's and 486's.

Thanks for your help.

Bob Ford & Jeff Martin
-- 
Bob Ford                            Voice:(814)234-8592 x36
INTERNET: astph!bob@cse.psu.edu	    FAX:  (814)234-1269
Philadelphia Phillies - 1993 National League Champions

-----------[000323][next][prev][last][first]----------------------------------------------------
Date:      Thu, 19 May 1994 14:35:59 GMT
From:      mefcs@coop.ch
To:        comp.os.os2.networking,comp.protocols.tcp-ip
Subject:   NFS with IBMs TCP/IP


Hello world,

I ran upon a problem installing an using NFS on two of our PCs.
In fact, installing was no problem but using is one.

I have an OS/2 V2.1 PC running IBMs TCP/IP V1.2.1 (I'm not sure
of the number. It is the newest version though) set up as an NFS
Server exporting Drive C (FAT) and D (HPFS). The second PC runs
DOS V6.0 with IBMs TCP/IP V2.1.1 (numbers ok, this is also the
newest version).

Problem is that on the DOS PC I can't start a Programm off a
mounted NFS drive no matter whether it is the OS/2 C or D Drive.
I can view text files (by means of TYPE) but again can't view
binary files.

Any clues ? Help would be much appreciated.

Felix

=================================================================
this ain't a signature...
all opinions stated are my own, other standard disclaimers apply
COOP Schweiz, Systemtechnik      email: mefcs@coop.ch
Felix Meschberger                   or: CS-DEP2.MEFCS@mvs.coop.ch
Freidorf 151                     phone: ++41 (0)61 336 55 42
CH-4132 Muttenz                  fax  : ++41 (0)61 336 53 12
=================================================================


-----------[000324][next][prev][last][first]----------------------------------------------------
Date:      19 May 94 11:35:54 +0200
From:      boris.benko@uni-mb.si (Boris B. Benko, MS)
To:        comp.protocols.tcp-ip,comp.sys.novell,comp.windows.x,comp.unix.misc
Subject:   Connecting PCs & wkstations...


	Hi all!

We've been trying to connect our two networks, token-ring and
ethernet network. We have a Novell server to connect our UNIX
based workstations on the ethernet side and PCs on the token-ring
side. We have a program named Chameleon, to access our workstations.
Well, all works fine, we are able to telnet or ping from PCs to
workstations (UnixWare). We would like to run X protocol as well, to
display stuff on the PC side. And there is a problem:
- if a PC is on the ethernet side (with workstations), all works fine
- but if a PC is on the token-ring side (and all network traffic goes
  through the LAN server), we can't display anything??? All the rest
  (telnet, etc...) works fine. We have some X tests (provided by
  Chameleon), which showed that some sort of X broadcasting is not
  possible. We are also able to send display from workstations to
  Chameleon. But we must do it from the workstation...

Did anybody experienced such a problem? Where seems to be a problem?
On workstations, with Chameleon or with the LAN server?

					Thanks in advance,
						BBB

-----------[000325][next][prev][last][first]----------------------------------------------------
Date:      19 May 94 14:09:54 CET
From:      s22562@seb.se (Bo Skogberg, S-E-B Data (+46 8 63)93433)
To:        comp.protocols.tcp-ip
Subject:   efnet irc server???

Hi,
Sorry for my unknowledge about things....
But what is an "efnet irc server"? Just some entries ago I found a list
of a lot of servers. But can anyone tell me what it is.
Thanks!
/Bo

-----------[000326][next][prev][last][first]----------------------------------------------------
Date:      Thu, 19 May 1994 15:25:56 GMT
From:      lagioia@clover.csata.it (Enzo Lagioia)
To:        comp.protocols.tcp-ip
Subject:   BootP and SLIP.


I'm in process of implementing some SLIP lines which connect a UNIX
workstation (as a server) to one or more PC's through PSTN.
I'd like to know whether there exists (like Cisco T/S) a version of SLIP package
(public domain) which supports Boot Protocol over asynchronous lines.

Thanks
Enzo Lagioia


-------------------------------------------------------
! Enzo Lagioia			Phone: +39 80 8770336 !
! Tecnopolis CSATA Novus Ortus	Fax  : +39 80 6951868 !
! Prov. Casamassima km.3			      !				
! 70010 VALENZANO (BA)  Italy			      !
!						      ! 
! E-Mail:	lagioia@csata.it	              !	
-------------------------------------------------------


-----------[000327][next][prev][last][first]----------------------------------------------------
Date:      19 May 1994 16:33:51 GMT
From:      zollner@iastate.edu
To:        comp.os.os2.networking,comp.protocols.tcp-ip
Subject:   Re: NFS with IBMs TCP/IP

In <Cq20K0.GyC@coop.ch>, mefcs@coop.ch writes:
>I have an OS/2 V2.1 PC running IBMs TCP/IP V1.2.1 (I'm not sure
>of the number. It is the newest version though) set up as an NFS
>Server exporting Drive C (FAT) and D (HPFS). The second PC runs
>DOS V6.0 with IBMs TCP/IP V2.1.1 (numbers ok, this is also the
>newest version).
>
>Problem is that on the DOS PC I can't start a Programm off a
>mounted NFS drive no matter whether it is the OS/2 C or D Drive.
>I can view text files (by means of TYPE) but again can't view
>binary files.

Felix: Here in the US, TCP/IP for OS/2 is at level 2.0. There is
a CSD at software.watson.ibm.com. If you indeed have 1.2.1 (look
at SYSLEVEL.NFS in \TCPIP) then you have an old version.

I have heard (but I am not sure) that different versions of IBM NFS
may use different buffer sizes. Try setting these buffers (see command
reference) to something like 4000 bytes and see if this helps.

************************************************************************
* Stefan Zollner, Assistant Professor and Associate Physicist          *
* Iowa State University/Ames Laboratory, A205 Physics, Ames, IA 50011  *
* zollner@iastate.edu    Phone: (515) 294-7327     FAX: (515) 294-0689 *
************************************************************************


-----------[000328][next][prev][last][first]----------------------------------------------------
Date:      Thu, 19 May 1994 16:36:20 GMT
From:      twcutter@mowgli.wr.usgs.gov (Tom W.Cutter)
To:        comp.protocols.tcp-ip
Subject:   Compression Modems & 56 Lease Lines?

Hi,
We have a number of 56K lines inter-connecting Cisco routers 
throughout the USGS.  I want to be able to NFS (automount) across these
slower lease lines without impacting interactive users.  I'm interested in
finding information on compression modems.  Anyone who is currently using
one, or any Recommendations of modems.  What's best!

I'm also interested in what NFS parameters any one has, that best works
across 56K lease lines.  I'm tried changing timeo to 1.4 seconds and
changed the read and write block size to 1K rather than 8K.  Any
suggestions would be appreciated.  The NFS is UDP, perhaps suggesting 
a NFS TCP Public Domanin implimentation.  The routers are using
TCP header compression.  Please respond in mail. I'll summaries.
Thanks in advance.

		- Tom	twcutter@mowgli.wr.usgs.gov	

-----------[000329][next][prev][last][first]----------------------------------------------------
Date:      19 May 1994 19:22:32 GMT
From:      jccw@fred.mitre.org (John C. C. White)
To:        comp.protocols.tcp-ip
Subject:   TCP Extensions Availability

We are looking at running TCP/IP on an ATM connection across
a couple of thousand miles. At OC-3 speeds we need the TCP
extension to allow a much larger window, per RFC-1323. I
suspect that patches to Sun's TCP to implement that change
are available somewhere, but I don't know where. Can anyone
point me to patches or executable which will let me use the
large TCP window?

-John White-
MITRE, Bedford MA


-----------[000330][next][prev][last][first]----------------------------------------------------
Date:      19 May 1994 16:35:12 +0200
From:      villa@ghost.sm.dsi.unimi.it (Max .Vilas. Villa)
To:        comp.protocols.tcp-ip
Subject:   IGRP proprietary

Hi, I need to know if the IGRP protocol (CISCO's routing protocol used 
in internet) is proprietary of CISCO or other manufacturers handle that.
Please send answers (if any) to villa@dsi.unimi.it.
Vilas
-- 
Massimo .Vilas. Villa (villa@dsi.unimi.it)
System and News Administrator
Computer Science Dep. - Milan University 

-----------[000331][next][prev][last][first]----------------------------------------------------
Date:      19 May 1994 20:15:59 GMT
From:      jtiller@ringer.jpl.nasa.gov (Jason Tiller)
To:        comp.protocols.tcp-ip
Subject:   Source for WU ftpd available?

Hello, all,

Is the source for WU ftpd (4.2 now?) available freely?  If so, where
would I find it?  If not, is another, similarly capable ftpd source
available for a porting project?

Thanks.

---
/============\   Jason Tiller  /======================================\
| Welcome to |                 | Voyager 2 emits less power than your |
| -> JPL <-/=======================\refrigerator light bulb - yet we  |
\==========|  Do *you* know where  |track it 6.4 *billion* km away!   |
   The     |  your spacecraft is?  |==================================/
     DSN   \=======================/ ** jtiller@ringer.jpl.nasa.gov **


-----------[000332][next][prev][last][first]----------------------------------------------------
Date:      Thu, 19 May 1994 20:20:20 GMT
From:      rbasket@cabell.vcu.edu (Robert E. Baskette)
To:        comp.protocols.tcp-ip
Subject:   PowerPC TCP/IP problem

HI,
  I have a simple but difficult problem.  I have four IBM RS/6000 PowerPC
units connected to a 10B-T hub.  These four units run AIX and were configured
for the network using the SMIT app.  Each of these four machines can
communicate with the world via our connection to the internet, BUT CANNOT
talk to each other.  The units are on their own sub-net connected to the
campus backbone by a Cisco IGS.  Telnet, ftp, ping work fine to any site open
on the internet.  Nothing is sent to one PPC from another PPC.  When a
connection is attempted, a DNS request is sent to the local nameserver.  The
nameserver resolves and responds.  After that no packets are sent from the
PPC.    
    
      Any ideas ???

Bob Baskette
Senior Computer Engineer
Virginia Commonwealth University

rbasket@cabell.vcu.edu



-----------[000333][next][prev][last][first]----------------------------------------------------
Date:      19 May 94 20:21:54 GMT
From:      sivapras@cs.iastate.edu (Gowri Sankar Sivaprasad)
To:        comp.protocols.tcp-ip
Subject:   CTRL-ALT-DEL problem with sockets betn. PC and Unix

Hello,

	I am facing a problem with UNIX sockets.
	
	Problem:  I establish a TCP connection between my SCO UNIX system and a
	Windows system (486 dx).  Then I reboot the windows m/c (CTRL_ALT_DEL).
	But the SCO system does not recognize that the client has gone down. 
	If I do a netstat on the SCO system, I find the socket to that client as
	"ESTABLISHED".  This is a major problem because it eats up TCP ports.
	
	My Windows application is in Visual Basic.

	Please suggest some solutions.

	PLEASE E-MAIL ME SINCE I DON'T READ THIS GROUP OFTEN.

	Thanks
	Gowri Sankar Sivaprasad



-----------[000334][next][prev][last][first]----------------------------------------------------
Date:      Thu, 19 May 1994 20:29:29 GMT
From:      gmz@netcom.com (Gerry Zeitlin)
To:        comp.protocols.tcp-ip
Subject:   Looking for "Router Engines"

I'm trying to find the phone number of a software development company
named Router Engines.  My understanding is that they OEM software to
router manufacturers to support the routing of TCP/IP and IPX packets.

Anyone know of them?
-- 


- Gerry Zeitlin / Oakland, CA / (510) 601-5758 -5760 FAX / gmz@netcom.com

-----------[000335][next][prev][last][first]----------------------------------------------------
Date:      19 May 1994 20:49:24 GMT
From:      rbr@dl.ac.uk (R.Bradshaw)
To:        comp.sys.sun.admin,comp.protocols.tcp-ip,comp.dcom.lans.ethernet
Subject:   Re: Two IP networks on one ethernet = broadcast storms

Jeff Makey (makey@VisiCom.COM) wrote:
: To ease the transition of several dozen hosts to new IP numbers, I
: want to support both the new and the old network numbers on the same
: ethernet wire.  To route IP packets between the old and the new, I
: have a SPARCstation 1 running SunOS 4.1.1 with 2 ethernet interfaces.
: It seems to work fine, except that a broadcast IP packet on the wire
: will turn into a broadcast storm.
 
:                            :: Jeff Makey
:                               makey@VisiCom.COM


I suggest you rebuild your kernel with the option DIRECTED BROADCAST
set to OFF in you config file. This simply means that an IP broadcast
packet received on the "other" interface won't be routed onto the
"correct" network. However, non-broadcast packets will continue to
be routed.

As an aside, if you have two interfaces onto the same cable, check
that they have different MAC addresses. As root, ifconfig them to 
check this out. If these are the same, it would be a good idea to 
change one of them; if one starts 8:0:20 then make the other 9:0:20...
The reason for this is that any ethernet frame sent to the 8:0:20...
mac address will be received on both interfaces. Luckily the IP
layer can handle duplicate datagrams :-)

Hope this helps.
-- 
Rob Bradshaw       Email: R.Bradshaw@dl.ac.uk       [rb685]
Tel:   +44 925 603226        |        Fax:   +44 925 603230
Networks Group, Daresbury Lab, Warrington, WA4 4AD, England

--
Rob Bradshaw       Email: R.Bradshaw@dl.ac.uk       [rb685]
Tel:   +44 925 603226        |        Fax:   +44 925 603230
Networks Group, Daresbury Lab, Warrington, WA4 4AD, England

-----------[000336][next][prev][last][first]----------------------------------------------------
Date:      19 May 1994 22:20:57 GMT
From:      jtiller@ringer.jpl.nasa.gov (Jason Tiller)
To:        comp.protocols.tcp-ip
Subject:   Re: Source for WU ftpd available? [Summary]

Hello, all,

Thanks to all the respondees about getting the source for wu-ftp 2.4.
I have been informed it is available in wuarchive, specifically:

wuarchive.wustl.edu:/packages/wu-ftpd/wu-ftpd-2.4.tar.Z

I've already gotten it and have begun looking it over (daunting, to
say the least!).  Thanks again for the quick replies.

---
/============\   Jason Tiller  /======================================\
| Welcome to |                 | Voyager 2 emits less power than your |
| -> JPL <-/=======================\refrigerator light bulb - yet we  |
\==========|  Do *you* know where  |track it 6.4 *billion* km away!   |
   The     |  your spacecraft is?  |==================================/
     DSN   \=======================/ ** jtiller@ringer.jpl.nasa.gov **


-----------[000337][next][prev][last][first]----------------------------------------------------
Date:      19 May 1994 22:29:25 GMT
From:      bill@msi.com (Bill Poitras)
To:        comp.sys.sgi.misc,comp.protocols.tcp-ip
Subject:   [Q] SLIP problems

I am connecting from a 486 computer with the Trumpet Winsock (a shareware 
TCP/IP kernel for Windows)  over SLIP to an Silicon Graphics Workstation 
running IRIX 4.0.5.  I can make the connection and telnet and FTP from 
the PC to the SGI I connect to.  However I can't connect outside of the 
machine to the rest of the network.  Here is the slip line the 
/usr/etc/remoteslip script I am using:

exec /usr/etc/slip -i -r $USER

I don't think I can install any hard coded routes on the SGI I am logging 
into, because it won't be the same one every time.  Basically I am 
logging into a terminal server, and then telnetting to my SGI of choice 
and logging in with a username identical to the hostname of my sliphost.  
It may sound weird, but it works great.

Will changing the slip line like this in the /usr/etc/remoteslip script 
help?

exec (route add host $USER localhost 0 ; \
      /usr/etc/slip -i -r $USER ; \
      route delete host $USER localhost)
 
I am hoping this will setup the necessary routing to allow traffic to 
escape the slip server host.  Before anyone says : "Well why don't you 
try it and find out", I have to give the system administrator (which is 
not me) solutions to the problem because his time is so tight.  He 
doesn't have time for too much trial and error.

If someone can just give reasons why this wouldn't work that would be 
find.  I don't understand routing very much.


--
+-------------------+----------------------------+------------------------+
| Bill Poitras      | Molecular Simulations Inc. | Tel (617)229-9800      |
| bill@msi.com      | Burlington, MA 01803-5297  | FAX (617)229-9899      |
 +-------------------+----------------------------+------------------------+
|FTP Mail           |mail ftpmail@decwrl.dec.com | Offers:ftp via email   |
|                   |Subject:<CR>help<CR>quit    |                        |
+-------------------------------------------------------------------------+

-----------[000338][next][prev][last][first]----------------------------------------------------
Date:      20 May 1994 11:54:13 -0700
From:      hvora@hsc.usc.edu (Heena Vora)
To:        comp.protocols.tcp-ip,comp.protocols.tcp-ip.domains
Subject:   Weird nameserver problems after IP address change with NIC

Hi all,

I'm helping some friends ser 

-----------[000339][next][prev][last][first]----------------------------------------------------
Date:      20 May 1994 12:04:20 -0700
From:      hvora@hsc.usc.edu (Heena Vora)
To:        comp.protocols.tcp-ip,comp.protocols.tcp-ip.domains
Subject:   Weird DNS/Nameserver problems after changing IP address with NIC

Hi all,
Ignore the earlier post (it was incomplete)

I'm helping some friends set up an internet node, and am encountering strange 
nameserver problems

We registered with the NIC and obtained an IP address of 198.147.76.1.  When we
actually connected through our service provider, this address  had to be subnetted
because the provider was short of network numbers.  So we became 198.147.76.65 on 
network 198.147.76.64.  

We informed the NIC and our secondary to make the change, which they did.  All this was 
about 10 days ago.  Since then, I've had trouble telnetting to a couple of 
nodes (such as usc.edu) for a random period of about 10 minutes or more 
during the day.  I'll get a message saying that the TELNET host has to close
the session because it can't determine our domain name from our IP address.

The NIC says that everything looks ok ontheir end, but they also said that
some root servers get updated rather slowly.  The thing I'd like to find out
is whether our old network number (ie 198.147.76.1) is still floating around there
and causing the problem.  Is it conceivable that the old and the new IP numbers
are updating each other in a infinite loop?  Or am I just being paranoid and
the problem will go away?

Thanks in advance!

Heena


-----------[000340][next][prev][last][first]----------------------------------------------------
Date:      20 May 1994 00:35:26 GMT
From:      bruce-b@cs.aukuni.ac.nz (Bruce Benson)
To:        comp.protocols.tcp-ip
Subject:   Problem with slip on Ultrix

Hi,

I am having trouble with SLIP on a DecSTATION 2100 running ultrix 4.3a.

I can make a connection and if I don't try and stuff too much data down
it it goes OK. However, when connected to a Mac running interslip over a
19.2k connection it freezes up if I try and download anything big (a
20k picture using Mosaic, for example).

netstat -m shows that a huge number of buffers have been allocated (200
rather than the usual 5-10). The only way of recovering is to reboot.

Anyone seen this before and got around it? Does anyone actually use
Dec's SLIP?

Thanks,
--
Bruce Benson.                            ,_o ,__o
Department of Computer Science.        _-\_\,\_<,
University of Auckland.               (*)/'--/'(*)
b_benson@cs.aukuni.ac.nz

-----------[000341][next][prev][last][first]----------------------------------------------------
Date:      Fri, 20 May 1994 00:58:26 GMT
From:      craigp@world.std.com (Craig Partridge)
To:        comp.protocols.tcp-ip
Subject:   funding for US students to attend SIGCOMM


*******************************************************************************

	    STUDENT GRANTS TO ATTEND ACM SIGCOMM '94

    ACM SIGCOMM '94 has received funding from the US National Science
Foundation (NSF) and Advanced Research Projects Agency (ARPA) to fund
up to eight graduate students of US universities to attend ACM SIGCOMM '94
in London (August 31 to September 2nd, with tutorials August 29 and 30).

The grants are up to $1000 each and intended to cover trans-Atlantic airfare,
hotel accommodations, and meals.  Students should note that the grant sizes
are fixed (any costs in excess of the grant will not be reimbursed).
SIGCOMM '94 will waive the conference fee for awardees.  Note that
all flights must be on a US carrier and be between the US and England to
be reimbursed, and that while students need not be US citizens or permanent
residents to apply, they do need a US Social Security number (for accounting
reasons).  Further note that SIGCOMM is only waiving conference fees, not
tutorial fees.

    Applications for grants will be evaluated by a three person committee
of Prof. Ian Akyildiz (Georgia Tech - chair), Prof. Lillian Cassel (Villanova)
and Prof. Craig Partridge (Stanford/BBN).  Grants will be awarded based
on the committee's estimate of how much attending the conference is likely
to benefit the student's graduate research, education and career prospects.
Some preference will be given to applicants from historically black colleges
and universities and minority institutions.

    To apply for a grant, a student should send a letter to Prof. Akyildiz
explaining how they believe attending the conference will benefit them.
The letter should be accompanied by a quote of the lowest available airfare
from the student's location to London and a letter from the student's graduate
advisor which evaluates the quality of the student's graduate work to date.
E-mail applications (to ian@armani.gatech.edu) are encouraged.  Applications
are due by June 10th.  The committee decisions will be made by June 25th.

    If the application cannot be e-mailed, send it via postal mail to:

	Dr. Ian F. Akyildiz
	Georgia Institute of Technology
	School of Electrical and Computer Engineering
	791 Atlantic Drive, NW
	Atlanta, GA 30332-0269

*******************************************************************************
*******************************************************************************

[Note to students -- I am aware of some less expensive hotels (c. 32 English
pounds per night) near the conference area, and am in fact going to stay in
one myself.  Contact me if you need recommendations -- Craig Partridge,
(craig@bbn.com).  Keep in mind that you don't need to give us a budget when
you apply -- just a plane fare quotation].


                           Advance Programme
                      ACM SIGCOMM'94 CONFERENCE
       Communications Architectures, Protocols and Applications

                      University College London
                              London, UK

                    August 31 to September 2, 1994
                 (Tutorials and Workshop, August 29-30)


                             Sponsored by
         The ACM Special Interest Group of Data Communication

This  conference provides an international  forum for the presentation
and discussion of communication network applications and technologies,
architectures,  protocols,  algorithms, and  performance models.   The
conference  and tutorials will be conducted on  the University College
London, London England.


		  ----------------------------------
		  T E C H N I C A L    P R O G R A M
		  ----------------------------------

Monday 29 August 1994

*  7:30AM - 5:00PM
   Tutorial and Conference Registration
   UCL CS Department, Pearson Building

*  9:00AM - 5:00PM, Tutorial T1
   "Personal Communication Services and Networks"
   Zygmunt Haas (AT&T Bell Labs)
   UCL CS Department, Pearson Building

*  9:00AM - 5:00PM, Tutorial T2
   "Protocol Performance"
   David D. Clark (MIT)
   UCL CS Department, Pearson Building

Tuesday 30 August 1994

*  7:30AM to 5:00PM
   Tutorial and Conference Registration
   Edward Lewis Lecture Theatre, Windeyer Building

*  9:00AM - 5:00PM, Workshop W1
   "Topics    in   High   Performance   Networking   Support   of
    Distributed Systems"
   Derek McAuley (University of Cambridge)
   UCL CS Department, Pearson Building

*  9:00AM - 5:00PM, Tutorial T3
   "Fiber Optic Networks"
   Paul  E. Green, Jr. (IBM Corporation)
   UCL CS Department, Pearson Building

*  9:00AM - 5:00PM, Tutorial T4
   "Multimedia Conferencing on the Internet"
   Van Jacobson (Lawrence Berkeley Laboratories)
   Edward Lewis Lecture Theatre, Windeyer Building

*  9:00AM - 5:00PM, Tutorial T5
   "Asynchronous Transfer Mode"
   Rainer Handel (Siemens Munich)
   UCL CS Department, Pearson Building

*  5:30PM - 8:30PM
   Welcoming Reception
   The Quad at University College London


Wednesday 31 August 1994

*  7:30AM to 5:00PM
   Conference Registration
   Edward Lewis Lecture Theatre, Windeyer Building

*  9:00AM - 10:00AM
   Session 1: Keynote Address
   (1994 ACM SIGCOMM Award Winner)
   Edward Lewis Lecture Theatre, Windeyer Building

*  10:30AM-12:30PM
   Session 2: Protocol Performance
   Experiences with a High-Speed Network

   Adaptor: A Software Perspective (Best Student Paper)
   P. Druschel (University of Arizona), L.L. Peterson (University of
   Arizona), & B.S. Davie (Bellcore)

   User-space Protocols Deliver  High Performance to Applications on a
   Low-Cost Gb/s LAN
   A. Edwards, G. Watson, J. Lumley, D. Banks,
   C. Calamvokis, & C. Dalton (Hewlett-Packard Labs, Bristol)

   TCP Vegas: New Techniques for Congestion Detection and Avoidance
   L.S. Brakmo, L.L. Peterson, & S.W. O'Malley (University of Arizona)

   A Structured TCP in Standard ML
   E. Biagioni (Carnegie Mellon University)

*  12:30PM  - 2:00PM
   Lunch

*  2:00PM-3:30PM
   Session 3: Congestion Management

   Making Greed Work  in Networks: A  Game-Theoretic  Analysis of
   Switch Service Disciplines
   S. Shenker (Xerox PARC)

   Scalable Feedback Control for Multicast Video Distribution in the Internet
   J. Bolot (INRIA), T. Turletti (INRIA) & I. Wakeman
   (University College, London)

   Statistical Analysis of Generalized Processor
   Sharing Scheduling Discipline
   Z.-L. Zhang, D. Towsley, & J. Kurose (University of Massachusetts)

*  4:00PM-5:30PM
   Session 4: ATM Flow Control

   The Dynamics of TCP Traffic over ATM Networks
   A. Romanow (Sun Microsystems) & S. Floyd (Lawrence Berkeley Labs)

   Reliable and Efficient Hop-by-Hop Flow Control
   C. Ozveren (DEC, Littleton), R. Simcoe (DEC, Littleton)  &
   G. Varghese (Washington University, St. Louis)

   Credit Update Protocol for Flow-Controlled ATM
   Networks: Statistical Multiplexing and Adaptive Credit Allocation
   H.T. Kung (Harvard University), T.  Blackwell (Harvard
   University), & A. Chapman (BNR)

*  7:30PM - 10:00PM
   SIGCOMM Social: Reception and Dinner
   The Dinosaur Room, Natural History Museum
   (Tickets  Required)


Thursday 1 September 1994

*  7:30AM to 5:00PM
   Conference Registration
   Edward Lewis Lecture Theatre, Windeyer Building

*  8:30AM - 10:00 AM
   Session 5: Internet Routing

   Flexible Routing and Addressing for a Next Generation IP
   P. Francis (NTT Software Labs) & R. Govindan (Bellcore)

   An Architecture for Wide-Area Multicast Routing
   S. Deering(Xerox PARC), D. Estrin (University of Southern
   California), D. Farinacci (Cisco Systems), V. Jacobson
   (Lawrence  Berkeley Labs), C.-G.  Liu (University of  Southern
   California) & L. Wei (University of Southern California)

   Distributed Routing Based on Link-State Vectors
   J. Behrens & J.J.  Garcia-Luna-Aceves (University of
   California at Santa Cruz)

*  10:30AM-12:00PM
   Session 6: ATM Switching and Signalling

   Signaling and Operating System Support for
   Native-Mode ATM Applications
   R. Sharma & S. Keshav (AT&T Bell Labs)

   Experiences of Building ATM Switches for the Local Area
   D.R. McAuley, R.J. Black & I.M. Leslie (University of Cambridge)

   Controlling Alternate Routing in General-Mesh
   Packet Flow Networks
   S. Sibal (RPI) & A. DeSimone (AT&T Bell Labs)

*  12:00PM  - 1:30PM
   Lunch

*  1:30PM-3:00PM
   Session 7: Nueral and Optical Networks

   On Optimization of Polling Policy Represented
   by Neural Network
   Y. Matumoto (I.T.S., Inc., Japan)

   An Optical Deflection Network
   J.  Feehrer  (University of  Colorado,  Boulder),  L.  Ramfelt
   (University of Colorado, Boulder/Royal Institute of Technology,
   Stockholm), & J. Sauer (University of Colorado, Boulder)

   Conflict-Free Channel Assignment for an Optical
   Cluster-Based Shuffle Network Configuration
   K.A. Aly (University of Central Florida)

*  3:30PM-5:30PM
   Session 8: Selected Topics

   MACAW: A Media Access Protocol for Wireless LANs
   V. Bharghavan (UC Berkeley), A. Demers (Xerox PARC),
   S. Shenker (Xerox PARC) & L. Zhang (Xerox PARC)

   Asymptotic Resource Consumption in Multicast
   Reservation Styles
   D.J. Mitzel (University of  Southern  California) & S. Shenker
   (Xerox PARC)

   Highly Dynamic  Destination-Sequenced  Distance-
   Vector Routing  (DSDV) for Mobile Computers
   C.E. Perkins & P. Bhagwat (IBM, Watson Research Center)

   A Methodology for Designing Communication Protocols
   G. Singh (Kansas State University)

*  5:30PM - 6:30PM
   SIGCOMM Business Meeting


Friday 2 September 1994

*  8:30AM - 10:00AM
   Session 9: Traffic Models

   Wide-Area Traffic: The Failure of Poisson Modeling
   V. Paxson & S. Floyd (Lawrence Berkeley Labs)

   Analysis, Modeling and Generation of Self-Similar
   VBR Video Traffic
   M.W. Garrett & W. Willinger (Bellcore)

   An Algorithm for Lossless Smoothing of MPEG Video
   S.S. Lam, S. Chow, & D. Yau (University of Texas, Austin)

*  10:30AM-12:00PM
   Session 10: Host Software
   USC: A Universal Stub Compiler
   S.W. O'Malley, T. Proebsting, & A. Montz (University of Arizona)

   An Object-based Approach to Protocol Software Implementation
   C.-S. Liu (Chung Yuan Christian University, Taiwan)

   Improved Algorithms for Synchronizing Computer Network Clocks
   D.L. Mills (University of Delaware)

*  12:00PM - 12:15PM
   Closing Session

Note: Program subject to change.

			  -----------------
			  T U T O R I A L S
			  -----------------

Tutorial T1
-----------
Zygmunt Haas, AT&T Bell Labs
"Personal Communication Services and Networks"

The  recent explosion  of interest  in  wireless and mobile  networks,
stimulated  by  the effort  of  Personal  Communication  Services  and
Networks (PCS & PCN) to  be  deployed at  the  beginning  of  the next
century,  suggests   the  enormous   technological,   scientific,  and
commercial potential in this field. The subject of wireless and mobile
communication  integrates  the  large body  of  knowledge  accumulated
through   the  traditional  radio   research,  the   large  networking
experience accumulated through the proliferation of LANs and WANs, and
the  vision  of  ubiquitous  connectivity anywhere,  at anytime,  with
anyone, and in any format.
   The  tutorial exposes both the theoretical  and  the practical
aspects  of mobile  networking,  from  a  networking  and  application
perspective.   We   will   present   the  concept,  architecture,  and
functionality of Personal Communications Services  and Networks (PCS &
PCN)  and  Universal  Personal  Telecommunications (UPT)  and  we will
describe the most  common  platform  for  mobile  communications:  the
wireless systems. In  particular, systems such  as cellular, cordless,
and satellite will be  discussed. Existing  and in-progress  standards
are also outlined.
   Finally, an abundance of examples of  the  wireless and mobile
networks will be described, giving realism to the presented material.
TOPICS:
* Elements of Wireless Mobile Communications
* Wireless Services and Applications
* The Cellular Concept
* The Cordless Concept
* Digital Communication Networks
* Local-Area Wireless Data Access
* Wide-area Wireless Data Access
* Mobile Satellite Communications
* Standardization of Wireless Networks
* PCS/PCN and UPT
* Summary: Where we have started and where are going .

Zygmunt Haas received his B.Sc. in EE in 1979 and M.Sc. in EE in 1985,
both with  Summa Cum Laude.   From  1979  till 1985  he worked for the
Government of Israel.  In  1988,  he  earned  his Ph.D.  from Stanford
University researching fast packet-switched networks, and subsequently
joined AT&T Bell Laboratories in Holmdel NJ, where he is now  a Member
of  Technical  Staff in the Wireless Networks Department.  Dr. Haas is
an author of  numerous  technical papers and holds several  patents in
the field  of high-speed networking, wireless  networks,  and  optical
switching.  He has organized  several workshops and  served as a guest
editor for  JSAC issues.  Dr. Haas is a Senior  Member of IEEE and his
interests  include:   mobile   and  wireless  communication  networks,
personal  communication services,  high-speed communication protocols,
and optical switching.



Tutorial T2
-----------
David D. Clark, MIT
"Protocol Performance"

Getting proper  performance from  a  network  or  protocol  is often a
difficult task. This tutorial uses examples from the Internet (TCP/IP)
protocol suite to illustrate critical performance issues. The focus is
on  providing  real-world  advice  on  how  to  design  and  implement
protocols in ways that  avoid performance  problems. The  presentation
will include  examples  of various  performance  problems and  how  to
detect and recognize them.

Topics
* Performance issues (reliability, throughput and delay)
* Implementation bottlenecks
* Specifications and their limitations
* Heterogeneity and its impact on implementation
* Network dynamics
* Visualizing protocol performance
* Limits of protocol performance

Dr. David Clark  is a senior research scientist  at MIT Laboratory for
Computer Science  and  a  recipient  of the ACM SIGCOMM  Award. He has
worked  on TCP/IP  since the  mid-1970s and  from  1981  to  1989  was
chairman of the Internet Activities Board. He is widely known for  his
insight into  protocol design and performance  and for  his  skill  in
identifying and  eliminating  myths about  protocol implementation and
performance.  His current areas  of  research include high-performance
networks,   the  evolution   of  the  Internet,  ATM  and  information
networking. He received his doctorate from MIT in 1973.

Tutorial T3
-----------
Paul E. Green, Jr., IBM
"Fiber Optic Networks"

Fiber  optic   technology  has  completely  transformed  the  internal
operation of the world's telephone networks and is beginning to impact
local  computer networks.  Compared to  the  voice  grade  phone  line
technology, which defined  most of the  network architectures  that we
are  still  living with  today,  fiber  offers ten orders of magnitude
better bandwidth and an  equal  improvement in  achievable  bit  error
rate.  By use of WDM and circuit switching, the additional benefits of
protocol transparency can be achieved.
   There is a widespread  feeling that  the generation of network
that  will  follow today's ATM and upgraded  Internet structures might
very  well   be  based   on  techniques  that  directly  unlock   this
revolutionary improvement at the physical level.
   The  course is  devoted  to  the  new  class  of "all-optical"
networks  that attempt  to  do  this.  The  lecturer  will  cover  the
optoelectronic components  involved and will  also treat  some  of the
network  architectural  consequences,  the  regulatory   and  economic
picture, and review some systems already implemented.

TOPICS
* Motivating fiber optic networks
* Fibers, couplers and taps
* Optical resonant structures
* Laser diodes and amplifiers
* Optical receivers
* System considerations
* Network topologies and link budgets
* Protocols, layers and network control
* Some implemented systems
* Status and prospects

Paul E. Green, Jr, is Manager of Advanced Optical Networking Member at
IBM Research in Hawthorne, NY.  He received the ScD degree from M.I.T.
in  1953, and after some  years at M.I.T. Lincoln Laboratory, where he
made pioneering  contributions to spread spectrum, adaptive receivers,
radar astronomy and seismic data processing, he joined IBM Research in
1969.  At  IBM  he has held  a  variety  of  management and  Corporate
Technical  Staff  positions. His technical interests  have centered on
computer  network  architecture,  and  he  has  received  several  IBM
Outstanding Innovation Awards for his  role in the initial formulation
and promotion of Advanced Peer to  Peer Networking,  now the basis for
further evolution of IBM's System Network Architecture. He is a member
of   the  National  Academy  of   Engineering,  in  1983   was   named
Distinguished Engineering Alumnus by North Carolina  State University,
and received the IEEE's Simon Ramo  Medal in 1991. He is the author of
many  technical   papers,  has   edited   several  books  on  computer
communications,  and  is  the  author  of  the  textbook  Fiber  Optic
Networks,  published  by  Prentice  Hall  in  June,1992.  He  has been
President of  both the IEEE  Communication  Information Theory Society
and the Communication Society.

Tutorial T4
-----------
Van Jacobson, LBL
"Multimedia Conferencing on the Internet"

An architectural overview  and detailed walk-through of  the protocols
and applications that provide  real-time, multiparty, audio, video and
shared workspace conferencing on today's Internet.
   Experiments and demonstrations over the Internet MBONE and the
DARTNET  testbed  have  shown   that   multimedia   and   conferencing
applications can indeed work  over  IP internets.  Playback algorithms
that  adapt  to  variations  in  network  delay  (such  as   VAT)  and
information  distribution  algorithms  designed  to  facilitate shared
workspaces (such  as those used in the  shared  whiteboard)  have made
these sorts  of applications  possible. This tutorial describes  these
algorithms and the applications that use them.
Topics
* IP as a real-time infrastructure: multicasting and queueing
* Adaptive Playback: VAT
* Managing Sessions: SD
* Managing Shared Workspaces: Shared Whiteboard
* Implications for the future of IP

Van Jacobson is a senior researcher at Lawrence Berkeley Laboratories,
where he works on real-time system performance, protocol and operating
system performance.  He is widely known for his groundbreaking work on
TCP/IP  performance,  TCP/IP   congestion  control,  and  support  for
multimedia  applications  on the  Internet. He is  the recipient of  a
number  of  awards and  teaches  periodically  at  U.C.  Berkeley  and
Stanford University.

Tutorial T5
-----------
Rainer Handel, Siemens Munich
"Asynchronous Transfer Mode"

The  tutorial  will  provide  a   comprehensive  introduction  to  the
Asynchronous Transfer Mode (ATM). Both technical and marketing aspects
of ATM will be addressed. ATM specification is not yet complete but in
a state that allows implementations which are basically compliant with
a worldwide agreed, unique standard supporting  data, voice, image and
multimedia applications.
       The presentation of the concept of ATM networking  will include
the ATM  protocol reference model,  the architecture of  ATM networks,
interfaces  and procotols, traffic  control  and resource  management,
signalling, operational  aspects,  ATM evolution  and  internetworking
aspects, and of course a detailed description of the ATM layer and ATM
adaptation layer  functions. An overview of how ATM cells are switched
and  transmitted  will  also  be  given. The  possible use of ATM in a
business   and  residential  environment  and  its  market  acceptance
depending on product availability, cost  and feature offerings will be
clarified.
TOPICS:
* High speed networks
* ATM concept
* ATM protocols
* ATM interfaces
* interworking and evolvability
* market  aspects
* switching and transmission products
* network  implementations and  service offerings

Rainer Handel  has  been with Siemens  (Public Communications Networks
Group)  since  1978 doing system  design and software  development for
switching  systems,  ATM  conceptual  and  standardization  work,  ATM
network and  product planning,  and currently long-term telecom market
and  technology trend evaluation. For several years  he  was active in
the standards bodies  CCITT, ETSI and T1, and is the author of several
papers and a book on ATM.

Workshop W1
----------
Derek McAuley, University of Cambridge
"Topics in High Performance Networking Support of Distributed Systems"

This  one day workshop will present the experiences of the speakers in
building various  components  of  distributed  systems  which  aim  to
effectively  utilise modern  high performance  networks. This workshop
consists of 4 talks. Each  talk will be 60 minutes with 15 minutes for
discussion.

1. The CHORUS Communication Architecture, Marc Rozier

The   communication   service  is   a  key  component  of  the  CHORUS
micro-kernel  architecture. First,  it provides  the  basic  framework
allowing  efficient   modular  operating  system  implementations.  By
dramatically reducing the overhead of  local communications, it is key
to the success of such  serverized implementations, which are now able
to  compete  with  monolithic  implementations.  Second,  it  provides
efficient, network-transparent, communication  services, well  adapted
to the distribution of  the operating system servers.   In particular,
it makes  possible  the  implementation  of UNIX systems  on massively
parallel architectures, offering a single system image to their users.
This tutorial will  address the  various aspects of this communication
architecture, from  the definition of the communication  services,  to
some  aspects of  its  implementation.  Emphasis  will  be  placed  on
insights from previous versions of this service.

2. The Organization of Networks in Plan 9, Rob Pike

In  a distributed system networks  are  of paramount importance.  This
tutorial   describes  the   implementation,   design  philosophy   and
organization of  network support  in  Plan  9. Topics include  network
requirements  for  distributed  systems,  our  kernel  implementation,
network naming, user interfaces and performance. We also observe  that
much of this organization is relevant to current systems.

3. Mixed media applications, David Tennenhouse

WWW  is a rapidly growing  phenomena which highlights the  interesting
applications possible with mixed media types. From experience with the
WWW this tutorial will  address the issues  raised in supporting these
mixed media types and the problems  in building systems  which support
media with time constraints.

4. What can you do with ATM today?, Derek McAuley

ATM must now be officially a  bandwagon. Some will tell you it  solves
all the world's problems because it was designed to, while others will
say  it's  good  for  nothing.  The  reality  and  hype  are  hard  to
distinguish. This talk will address what ATM can be used for today and
highlight those features for which it  is rightly criticised not least
of  which is  end-system  integration.  The talk could  be  subtitled,
"Difficult questions to ask your ATM salesman''.


Marc Rozier is the head  of the Micro-Kernel Department  within Chorus
systemes. He graduated from Ecole Nationale Superieure Informatique et
de  Mathematiques Applique'es  de Grenoble (ENSIMAG) before earning  a
doctor's   degree   in  Computer   Science  from   Institut   National
Polytechnique  de Grenoble (INPG). In 1981-82,  he was involved in the
CESAR  project at  IMAG  (Grenoble),  working  on  the  Validation  of
Distributed Systems. He gained experience in programming languages for
distributed applications  and distributed  systems. He joined INRIA in
1982 as  a  researcher  in  the CHORUS  distributed  operating  system
project. In 1987, he became one of the founders of Chorus systemes. He
is one of the main designers of the CHORUS-v3 Micro-Kernel technology.
He is the author of several publications in international journals and
conferences.

Rob Pike is  well known for his appearances on "Late Night with  David
Letterman",  is  also  a  Member  of  Technical  Staff  at  AT&T  Bell
Laboratories in Murray Hill, New Jersey, where he has been since 1980,
the same year he won  the  Olympic silver medal in Archery. In 1981 he
wrote the first  bitmap window system  for Unix systems, and has since
written ten  more.  With Bart Locanthi he designed  the Blit terminal;
with Brian Kernighan he  wrote The Unix  Program- ming Environment.  A
shuttle mission nearly  launched a gamma-ray telescope he designed. He
is a Canadian citizen and has never written a program that uses cursor
addressing.

David Tennenhouse is  an Assistant  Professor of Computer Science  and
Electrical Engineering at MIT's Laboratory for Computer Science. He is
leader  of  the  Telemedia,  Networks  and  Systems  Group,  which  is
addressing  systems  issues  arising  at   the  confluence   of  three
intertwined  technologies: broadband networks,  high  definition video
and distributed computing.   David  studied electrical  engineering at
the  University of Toronto, where he received  his B.A.Sc. and M.A.Sc.
degrees.  In 1989 he completed his Ph.D. at the Computer Laboratory of
the University  of Cambridge. His Ph.D.  research focused on ATM-based
site interconnection issues. This work, which was conducted within the
Unison  Project, led to  the early implementation of an ATM-based wide
area testbed.

Derek  McAuley  is  a  Lecturer  in  the  Computer  Laboratory at  the
University  of  Cambridge. His research  interests include networking,
distributed   systems   and   operating  systems.    Recent  work  has
concentrated on the support  of time dependent  mixed  media types  in
both  networks and operating systems. He has failed to leave Cambridge
since  arriving in  1979 to read Mathematics. In 1989 he completed his
Ph.D. on ATM internetworking. He has had a hand in  de-commissioning 4
ATM  networks,  including  Tennenhouse's carefully  constructed Unison
platform.


			   ---------------
			   L o c a t i o n
			   ---------------

The conference will be held in the Edward Lewis  Lecture Theatre which
is located in the  Windeyer Building on the UCL campus.  This building
is located on the corner of  Cleveland Street and Howland Street, with
the entrance  on  Cleveland Street.  Tutorials are all in UCL Computer
Science  Department in the Pearson Building, except T4 (Van  Jacobson)
on  the Tuesday which is held in the Edward Lewis Lecture Theatre.

The main entrance of UCL  is located at the north end of Gower Street,
close to Euston Square, Warren Street,  or Euston tube stations.   The
UCL  Computer Science  Department is located in  the  basement of  the
Pearson Building.  Location


		     ---------------------------
		     T r a n s p o r t a t i o n
		     ---------------------------

* Getting to  London

There  are  four  airports  in  and  around   London.   Here  is  some
information that  might help you to plan your journey.  Please consult
your travel agency or the airports directly for further information.

LONDON Heathrow Airport: 24 km west of London
Telephone: +44-81-745-6156
LONDON Gatwick Airport: 46 km south of London
Telephone: +44-293-535-353
STANsted Airport: 55 km north east of London
Telephone: +44-279-680-500

* Getting to UCL and Hotels

UCL is located  in central London, and is served by Warren St,  Euston
and Euston Square Underground (tube) stations, as well as several main
bus  routes.   The department  of  computer  science is  right  by the
entrance to the main quadrangles, on Gower Street.

>From Heathrow:  Best  by  tube with Victoria Line  to  Euston  Station
(about #3, 50 minutes). Alternatives are via Bus with London Transport
A1 Airbus to Victoria Station (45 minutes).

For local hotels it is probably best to go to Euston Station and get a
taxi from  there  unless  you have a  street map already  and know the
nearest tube station.  A free tube map may  be obtained at any  ticket
office.

>From Gatwick:  Best  by train, BR  Gatwick Express  to London Victoria
Station every 15 minutes (about #8.60, 30 minutes).

Unless you plan to sightsee  outside London a car  is probably a waste
of time.  Tube  fares are based on a zone system. After 9:30AM you can
get One Day Travel cards which allow you unlimited travel within given
zones  for the rest of the day -  that includes train and bus services
within  that zone too.  Zones 1,2 & 3 #2.30 pounds.   Zones  1-5 #2.60
pounds.


		      -------------------------
		      A c c o m o d a t i o n s
		      -------------------------

The following hotels are  walking distance from the conference meeting
room  on  the  UCL  campus.   Contact  the  hotel  directly  to  place
reservations.It  is highly recommended that reservations  are  made as
early as possible. Refer to SIGCOMM'94 when making the reservation.

* 	Hotel Ibis Euston
	3 Cardington Street, NW1
	Telephone: +44-71-388-7777, Fax: +44-71-388-0001
	Total Rooms: 300
	Single Room #49.50, Double Room #49.50
	Near UCL, about 10 minute walk from main Conference Hall.

* 	St. George's Hotel
	Langham Place, W1N
	Telephone: +44-71-580-0111, Fax: +44-71-436-7997
	Total Rooms: 86
	Single Room: #80.00, Double Room: #100.00
	(Includes Continental Breakfast)
	Situated near Oxford Circus, about 10 minute walk from main venue.

* 	RAMSAY HALL
	20 Maple Street, W1P
	Total Rooms: 400
	Telephone: +44-71-387-4537, Fax: +44-71-383-0843
	Single Room: #19.50, Double Room: not available.
	(Includes Continental Breakfast)
	Student residence used as hotel during summer break, 5 minute walk
	from main conference venue.

* 	Hotel Russell
	Russell Square, WC1
	Telephone: +44-71-837-6470, Fax: +44-71-837-2857
	Total Rooms: 328
	Single Room: #70.00, Double Room: #90.00
	(Includes Continental Breakfast)
	Old Victorian Style Hotel. About 15 minute walk from Conference
	venues. Russel Square  Station is on the Picadilly  line which
        reaches	to Heathrow Airport. Airport Bus stop nearby as well.

* 	Forte Crest Bloomsbury
	Coram Street, WC1
	Telephone: +44-71-837-1200
	Fax: +44-71-837-5374
	Total Rooms: 230
	Single Room: #69.00, Double Room: #79.00
	(Includes Continental Breakfast)
	Modern hotel near Hotel Russell.

There  are  a large  number  of hotels near the conference. Almost any
hotel in the WC1 area of London is within 15 minutes walking distance.
A    list    of    more    hotels    may    be     found    via    www
(http://www.cs.ucl.ac.uk/sigcomm94)       or       anonymous       ftp
(norman.eng.buffalo.edu:/pub/SIGCOMM94). The list also includes nearby
lower cost housing and youth hostels.


	   ------------------------------------------------
	   R E G I S T R A T I O N    I N F O R M A T I O N
	   ------------------------------------------------

Full conference  registration  includes breaks, lunch, Tuesday evening
reception, one  ticket to dinner in the  Dinosaur Room of the  Natural
History Museum on Wednesday, and a copy of the conference proceedings.

Student  registration includes breaks, lunch and proceedings but  does
not include the  dinner/museum event.  On site registration will begin
Monday August 29, 1994 from 7:30AM - 5:00PM, and every day of the con-
ference starting at 7:30 am.



ACM and SIGcomm Membership
--------------------------

If  you are not an ACM or a SIGCOMM member at this time, you may  join
now to take  full advantage of ACM/SIGcomm Member or Student rates for
SIGCOMM94:

- ACM Associate Member Dues      	$82/#52
- Add SIGCOMM to ACM Membership      	$22/#15
- ACM Student Dues                  	$25/#17
- Add SIGCOMM to ACM Student Membership	$15/#10
- SIGCOMM Membership only (non-ACM)  	$50/#32

Total Membership Fees              $/#  _________

(Note: $ indicates U.S. dollars, and # British Pounds Sterling)

To advance the sciences and arts of information processing; to promote
the free interchange of  information about  the sciences  and  arts of
information processing both among  specialists and  among the  public;
and  to  develop  and  maintain  the  integrity  and   competence   of
individuals engaged  in  the  practice  of  information processing.  I
hereby affirm that I subscribe  to the purpose of  ACM and  understand
that my membership is not transferable.

Signature _________________________________________ Date ____________


Tutorials
---------

Check each tutorial attending.  The tutorial registration fee includes
one copy  of the tutorial notes and  lunch.  Tutorials  are on a first
come first serve basis.

- T1 	Personal Communication Services & Networks (Monday)
- T2 	Protocol Performance (Monday)
- T3 	Fiber Optic Networks (Tuesday)
- T4 	Multimedia Conferencing on the Internet (Tuesday)
- T5 	Asynchronous Transfer Mode (Tuesday)
- W1 	Workshop on Distributed Systems (Tuesday)

Tutorial Rates
			Postmarked by         	Postmarked
			aug/1/1994           	after aug/1/1994

ACM/SIG Member          _____@ $275/#172        _____@ $325/#205
Non-Member              _____@ $350/#220        _____@ $400/#250
Student                 _____@ $138/#87         _____@ $175/#110

Total Tutorial Fees     _____$/#                _____$/#


Special Needs
-------------

Vegetarian Meals:    	- Yes  	- No


Conference Registration
-----------------------

Please complete and  send  registration form, with check,  credit card
information or money orders (no purchase orders) to the address below.
Registrations accepted  via postal  mail,  fax or  email (with  credit
card) only.

				Postmarked by         	Postmarked
				Aug/1/1994           	after Aug/1/1994

ACM/SIG Member			_____@ $315/#200 	_____@ $365/#230
Non-Member     			_____@ $397/#252  	_____@ $440/#275
Student       	 		_____@ $100/#63 	_____@ $130/#82

Total Registration  Fees    $/# _____               $/# _____

Extra Dinner/Museum Ticket      _____@ $55/#35


TOTAL ENCLOSED             $/#  _____ (ACM/SIGCOMM Membership, tutorials,
                                       conference registration)



NAME _________________________________________________________________

AFFILIATION __________________________________________________________

ADDRESS ______________________________________________________________

______________________________________________________________________

PHONE _____________________________ FAX ______________________________

Email ADDRESS ________________________________________________________

SIGCOMM Member? 	- YES    	 - NO
ACM/SIGCOMM Member Number ____________________________________________

CREDIT CARD PAYMENT 	- VISA   - MASTERCARD   - euroCARD

CARD HOLDER NAME _____________________________________________________

CARD NUMBER ______________________________________ EXP. DATE _________

SIGNATURE ____________________________________________________________


Please send this  form and a check,  credit card information  or money
orders (no purchase orders) to SIGCOMM'94.  Registrations accepted via
postal mail, fax or email only.

Send U.S. or   				Send Pound Sterling
Credit Card Payments to:

Patrick McCarren                        Soren-Aksel Sorensen
ACM - 17th Floor                        Dept. of Computer Science
1515 Broadway                           University College London
New York, NY 10036                      London WC1E 6BT
USA                                     United Kingdom
phone: +1 212/626/0611                  phone: +44 71 380 7269
fax: +1 212/302-5826                    fax +44 71 387 1397
mccarren@acm.org

Email registrations  can  only  be  made  by a credit card  during the
pre-registration period ending 1 August 1994 and must use credit  card
payment.   A registration confirmation  letter  will  be sent to  each
participant  upon  receipt of  the  completed  registration  form  and
accompanying  payment.   Registration fee  will  be  refunded,  less a
$30/#19 administration fee, if  cancelation  notification  is received
prior  to  15  August  1994.  Substitution  for  a  paid  attendee  is
acceptable.

	    ----------------------------------------------
	    C o n f e r e n c e    O r g a n i z a t i o n
	    ----------------------------------------------

General Chair:   Jon Crowcroft, University College London
Program Chairs:  Stephen Pink, Swedish Institute of Computer Science
                 Craig Partridge, BBN (Program Co-Chair for North America)

Ian F. Akyildiz, Georgia Institute of Technology, USA
Lillian N. Cassel, Villanova Univ., USA
Vinton Cerf, MCI, USA
Lyman Chapin, BBN, USA
Jon Crowcroft, Univ. College London, UK
Andre Danthine, Univ. of Liege, Belgium
Gary Delp, IBM, USA
Patrick W. Dowd, SUNY/Buffalo, USA
Deborah Estrin, Univ. Southern California, USA
David Feldmeier, Bellcore, USA
Sally Floyd, Lawrence Berkeley Laboratory, USA
David Greaves, ORL Cambridge, UK
Per Gunningberg, Swedish Institute of Computer Science, Sweden
Christian Huitema, INRIA, France
David Hutchison, Lancaster Univ., UK
Raj Jain, Ohio State University, USA
Jim Kurose, Univ. of Massachusetts, USA
Ian Leslie, Univ. of Cambridge, UK
David Oran, Digital Equipment Corp, USA
Gerard Parr, University of Ulster, Northern Ireland
Guru Parulkar, Washington Univ. St Louis, USA
Krzysztof Pawlikowski, Univ. of Canterbury, New Zealand
Bernhard Plattner, ETH Zurich, Switzerland
Scott Shenker, XEROX PARC, USA
Deepinder Sidhu, Univ. of Maryland-BC, USA
Jonathan M. Smith, Univ. Pennsylvania, USA
Khosrow Sohraby, Univ. of Missouri - Kansas City, USA
James Sterbenz, IBM Research, USA
Greg Watson, Hewlett Packard Labs, UK
Greg Wetzel, AT&T Bell Laboratories, USA
Lixia Zhang, XEROX PARC, USA


	 ---------------------------------------------------
	 F O R   A D D I T I O N A L   I N F O R M A T I O N
	 ---------------------------------------------------

Additional information may be found/requested from:

www:           http://www.cs.ucl.ac.uk/sigcomm94
anonymous ftp: norman.eng.buffalo.edu:/pub/sigcomm94
email:         sigcomm94@eng.buffalo.edu
fax:           +1 716.645.3656
phone:         +1 716.645.2406

-----------[000342][next][prev][last][first]----------------------------------------------------
Date:      Fri, 20 May 1994 01:33:18 GMT
From:      abell@velveeta.apdev.cs.mci.com (Andrew_Bell)
To:        comp.protocols.tcp-ip
Subject:   Re: How does a server detect a client crash using sockets?

In article <1994May10.194754.16641@noao.edu> rstevens@noao.edu (W. Richard Stevens) writes:
>> As a followup, how does so_keepalive work?
>
>Check out my recent book "TCP/IP Illustrated" (Addison-Wesley, 1994).

A wonderful book.  Buy it.  What topics might we expect to see
covered in Volume II?

>There's a complete chapter with all the gory details of the keepalive
>option.  Also check out Appendix E of this book for common configuration
>parameters for the keepalive option, since everyone's first question
>after learning what it does is "how can I set the keepalive time to
>some value less than 2 hours."
>
>> I set this option
>> on the socket and the documentation says that a SIGPIPE should be returned
>> to the process associsted with the process, but I do not seem to get it after
>> the client has crashed. Does my server application have to send a periodic
>> transmission on the socket, or is that handled by the lower layers?
>> 	Any and all responses appreciated. I am running this application
>> on an IBM RS6000 running AIX 3.2.4
>
>Your documentation is basically wrong.  TCP generates SIGPIPE when an
>RST is received on a connection--you're writing to a connection that's
>been reset by the other end.  TCP does *not* generate this when your
>end receives a FIN, due to TCP's half-close feature.
>
>How to detect that the client has "crashed" requires that you be more
>specific with what "crash" means.  If the client *process* crashes, its
>TCP should send a FIN, which delivers an EOF to your end of the connection.
>If the client *host* crashes, a keepalive probe can detect this, though
>the error returned to your end of the connection depends whether the
>client host has rebooted or not when the keepalive probes are sent.  If
>the client *host* crashes and the connection is idle (you don't try to
>send something to the client) and the keepalive option is not set, your
>end will sit there, and sit there, and sit there ...
>
This I understand.  My question is what happens to things in buffers
when a process goes down?  Certainly the write() call will succeed even
if a read() has not yet occurred at the other end.  Is there a way to know
that the process at the other end of a socket connection has read the
data that TCP has sent?  Is there any way, within TCP, to know that the
process at the other end of a socket has crashed and left unread stuff
in a buffer?  How can the sending process know what stuff never got read
and needs to be retransmitted when a new connection is made?  Does this
have to be done within the application (client and server)?  Does any
UNIX OS with asynchronous I/O help the situation (write now and tell me
when the read happened later)?

Answers to any of the above are appreciated.

Andrew Bell
MCI Communications
abell@chong.nms.cs.mci.com

-----------[000343][next][prev][last][first]----------------------------------------------------
Date:      20 May 1994 01:50:53 -0000
From:      vbraco@news.delphi.com (VBRACO@DELPHI.COM)
To:        comp.protocols.tcp-ip
Subject:   UDP SOCKET HELP

I'm trying to write a program that will send data out on a socket which
can be read by one or more programs on other hosts. The hosts are HP's.
The HP documentation has led me to believe that there is a way to
'broadcast' this data in such a way that it can be received by multiple
hosts simultaneously. I have a pair of simple prototype programs which
communicate using send() & recv() but the 'receiver' only seems to get the
message when the 'sender' does a connect() before the send() or if
sendto() is used. In each case the IP address of the host running the
'receiver' must be explicitly specified or the 'receiver' doesn't see the
message. I have tried specifying the wildcard address and using the
setsockopt() function to set SO_BROADCAST but so far - no joy.  No
response from the 'receiver' is desired, however, there may be several
'receiver' processes on different hosts all wishing to read the 'sender's
data.  Anybody know how to do this? A code fragment would be very much
appreciated. If this is documented anywhere, please take a moment to point
me at it.

Thanx-In-Advance

Vince





-----------[000344][next][prev][last][first]----------------------------------------------------
Date:      Fri, 20 May 1994 09:21:40
From:      clarkb@netstar.com (Clark Bremer)
To:        comp.protocols.tcp-ip
Subject:   Re: Forwarding a socket connection to another machine

In article <2r8ivfINNhk9@early-bird.think.com> barmar@think.com (Barry Margolin) writes:
>From: barmar@think.com (Barry Margolin)
>Subject: Re: Forwarding a socket connection to another machine
>Date: 16 May 1994 19:53:19 GMT
 
>In article <TBEL.94May13214336@oahu.cern.ch> tbel@oahu.cern.ch (Tim Bell)
>writes:
>>Is there a way to forward a socket connection to another machine ? I
>>was to write a gateway program that will get called when a connection
>>is made but then forward on the connection to another machine (it
>>won't be the same machine each time).  Since a large amount of data
>>will be transferred, I don't want to put a load on the intermediate
>>machine so a simple read/write forwarder won't work... in pictures...
 
>There's no built-in mechanism in TCP/IP for this.  Some application
>protocols have facilities for this; for instance, FTP has a command that
>can be used to tell one server to send a file to another server, rather
>that pulling the file to the client and then sending to the second server.
 
>>     A  ------->   B
>>                       ---------> C
>>
>>     A --------------------------> C
 
>In this case, it seems most straightforward for the A-B connection to start
>off with B sending back C's address.  Then A would close the connection
>with B and open a new connection with C.
>-- 

I've seen a variation of this done with a name server, to implement load 
balance.  The user telnets to, say, 'machine.com', and doesn't realize that 
he/she is really being connected to one of 10 machines (like 'machine1.com' 
thru 'machine10.com').  When the client looks up 'machine.com', the hacked-up 
name server returns the ip address of whichever machine has the least load at 
that time.  The client then makes a connection to the correct machine, 
whithout knowing that it could be any of the 10.  How the name server figured 
out which machine was least loaded was a proprietary scheme implemented with 
UDP, but this could probably have been done with SNMP.  CB.
===========================================================================
          _  _               Clark Bremer     clarkb@netstar.com
         /  /_)              Software Engineer, NetStar Inc.
         \_/__)              10250 Valley View Road  MPLS, MN 55344

-----------[000345][next][prev][last][first]----------------------------------------------------
Date:      20 May 1994 08:07:22 GMT
From:      barrya@barrypc.sct (GodMachine)
To:        comp.protocols.tcp-ip,comp.protocols.tcp-ip.domains,comp.protocols.tcp-ipibmpc
Subject:   problem in setting up sub-network in super-tcp/ip

Hello, everyone! I am working in a company in which, only some of domains has ftp access to 
the outside world. The domain I am networked has no ftp access to the outside world, only local
access. I am currently using super-tcp/ip in my pc. I noticed that you can define a sub-network
in super-tcp/ip. My question is:-

I have access to one of the sparc workstation which is networked with another domain, and it
has ftp access to the outside world. Can I configure my super-tcp/ip, so that the sparc station
will be my sub-network. Thus I can ftp out via this sub-network?

Thanks in advance. 

BTW, individual reply is welcome, since I have no internet email address, you guys can reply  to
auyeung_b@bt-web.bt.co.uk.

Barry.

-----------[000346][next][prev][last][first]----------------------------------------------------
Date:      20 May 94 13:52:06
From:      drw@nevanlinna.mit.edu (Dale R. Worley)
To:        comp.protocols.tcp-ip
Subject:   Re: Resale of registered

In article <1.6133.2441.0N27CADA@dscmail.com> john.will@dscmail.com (John Will) writes:
   D >*Domain names*, on the other hand, are likely to be worth money,
   D >because they aren't interchangable.

   Why would a domain name be worth anything?  I can register as many 
   domain names as I can invent, doesn't seem to be a commodity...

Go over to comp.org.eff.talk and ask "Why does MTV want to own
MTV.COM."  (Right now, the company that owns MTV is suing the fellow
who has registered MTV.COM.)

The reason is not that names are scarce, but that names *mean
something* in the minds of the customers.  Companies spend thousands
of dollars, and sometimes millions, picking names for things.

Dale

Dale Worley		Dept. of Math., MIT		drw@math.mit.edu
--
Have you heard about "neutron buildings"?  The investors are destroyed, but
the buildings remain standing.
-- Timothy C. May

-----------[000347][next][prev][last][first]----------------------------------------------------
Date:      20 May 94 13:53:46
From:      drw@nevanlinna.mit.edu (Dale R. Worley)
To:        comp.protocols.tcp-ip
Subject:   Compression in DSUs

On page 41 of the 16 May 94 Network World there is an article on a
standard for synchronous data compression in DSUs, and its progress
through the standardization maze and in the market.  If you're
interested in the topic, it's probably useful.

Dale

Dale Worley		Dept. of Math., MIT		drw@math.mit.edu
--
Clinton's claim that he smoked pot but didn't inhale is an insult to
dope fiends everywhere.	-- Hunter S. Thompson

-----------[000348][next][prev][last][first]----------------------------------------------------
Date:      20 May 1994 09:06:18 GMT
From:      tom@uni-paderborn.de (Torsten Metzner)
To:        comp.protocols.tcp-ip
Subject:   tcp/ip C++ class library

Hello,
I am interested if there are some Tcp/IP C++ class libraries available
for different Operating Systems. 

The Operating Systems I am mostly interested are:
   Unix in general, especially:
      SunOS 4.1.a
      Solaris 2.b
      AIX 3.c
      IRIX 4.d
      IRIX 5.e
      LINUX
   Macintosh System 7.f

Any help is greatly appreciated.

Thanx a lot in advance,

    Torsten.

    E-Mail:   tom@uni-paderborn.de


-----------[000349][next][prev][last][first]----------------------------------------------------
Date:      Fri, 20 May 1994 09:28:57 GMT
From:      denis <denis@bfh.irkutsk.su>
To:        comp.protocols.tcp-ip
Subject:   Помогите восстановить данные!!!

        Помогите выйти из ужасной ситуации!

   Дело в том, что при работе в NetWare 3.11 на файл-сервере EverEX Step-VL
стали возникать проблемы с синхронизацией дисков (при отражении). При этом
резко упала производитеельность в сети (именно по этому мы не завершили
работу SBackup'а - как на зло!).
  Мы (- @#$%^&) решили пока выключить "зеркалку" (запустили Install). Но по
полному незнанию особенностей подлого новеля, не перегрузив сервер, грохнули
всю информацию о томах на одном диске (а что делать, если нетваре ругалась,
что тома на дисках одинаковые?). Ну и, естественно тоже самое отразилось и на
второй диск. В итоге мы остались без чрезвычайно важных данных (я уже не
говорю о куче документации, безвременно почившей с ними - бог с ней).
  Больше никаких действий с дисками мы не производили (убрали в сторонку).

  И вот, привыкшие к возможностям МС-Доса, мы не можем не верить в возможность
восстановления данных.
  ТАК ЛИ ЭТО? Может есть какой-нибудь софт? Или хотя бы можно узнать, в каком
месте на винте новель держит (-ал) данные о томах?

  P.S.  На всякий случай: винт - Seagate ST3600N.
                                                     Денис Голубев


-----------[000350][next][prev][last][first]----------------------------------------------------
Date:      20 May 94 18:06:27 -0400
From:      sysmgr@wittenberg.edu (Sesquipedalian)
To:        comp.protocols.tcp-ip
Subject:   Re: PowerPC TCP/IP problem

In article <1994May19.202020.3744@cabell.vcu.edu>, rbasket@cabell.vcu.edu (Robert E. Baskette) writes:
> HI,
>   I have a simple but difficult problem.  I have four IBM RS/6000 PowerPC
> units connected to a 10B-T hub.  These four units run AIX and were configured
> for the network using the SMIT app.  Each of these four machines can
> communicate with the world via our connection to the internet, BUT CANNOT
> talk to each other.  The units are on their own sub-net connected to the
> campus backbone by a Cisco IGS.  Telnet, ftp, ping work fine to any site open
> on the internet.  Nothing is sent to one PPC from another PPC.  When a
> connection is attempted, a DNS request is sent to the local nameserver.  The
> nameserver resolves and responds.  After that no packets are sent from the
> PPC.    
>     
>       Any ideas ???
> 
> Bob Baskette
> Senior Computer Engineer
> Virginia Commonwealth University
> 
> rbasket@cabell.vcu.edu
> 
> 
Is the DNS server on the same subnet as the AIX machines?  If not, then maybe
the AIX machines don't have the correct subnet mask?  It sounds like as long as 
they have to forward to the router, everything is ok.  You might take a look at
the AIX machine's routing table (netstat -r) and see where it thinks it should
be sending things on the local subnet.  It should point to it's own interface.
I don't know how good smit is, but the admin tool with our NCR set things up
mostly right, but I ended up having to set up the routing by hand (it assumed I
was running routed).

Just a thought.

Kevin Yoders
System and Network Manager
Witteberg University

-----------[000351][next][prev][last][first]----------------------------------------------------
Date:      Fri, 20 May 1994 12:35:29 GMT
From:      wku@rci.ripco.com (Werner Kubitsch)
To:        comp.protocols.tcp-ip
Subject:   Wanted, software for high availablility, redundancy

Hi everybody,

We are designing a system, that involves a unix system talking to an IBM 
mainframe via LU6.2 and to a number (up to 25) of PC's via TCP/IP. Now 
we need to implement high availability by having multiple UNIX systems
keeping redundant information. So all the information from the mainframe 
as well as the PC's needs to be mirrored onto all the UNIX systems. Also
if the primary UNIX system fails, the PC's need to switch to one of the
mirrored systems, making this the primary then.

So far we've been looking into a product called ISIS, which provides some 
of that functionality. I'm interested if anybody knows of any other products
that might help us in creating this system. 

Any input would be greatly appreciated.

-- Werner

-----------[000352][next][prev][last][first]----------------------------------------------------
Date:      Fri, 20 May 1994 13:35:35 GMT
From:      ccksb@dakhma.trentu.ca (Ken Brown)
To:        comp.protocols.appletalk,comp.protocols.tcp-ip,comp.sys.mac.comm,fr.network.divers
Subject:   Re: MacTCP 2.0.4 & BOOTP RFC 1048 & DNS


[description of bootp problems snipped...]

The way I found to deal with this was to use a NetWare 3.xx server running
HellSoft's bootpd and resolv.  If I set up MacTCP to use a server and specify
the DNS info as below the Mac will a) get an ip, b) get the nameserver and 
c) get the gateway - all correctly.

DNS stuff

 default.domain    xxx.xxx.xxx.xxx     x
 .                 xxx.xxx.xxx.xxx     o
 .                 yyy.yyy.yyy.yyy     o

Hellsoft's bootpd can dynamically or statically serve ip addresses and can
exclude hosts - al via config files.  I had to specifically exclude hosts
I didn't want bootp served from the NW host (eg: NCD X-terminals).

You'll find bootpd18.zip and resolv13.zip on blaze.trentu.ca in 
~ftp/pub/netware.

This description might assist in your use of HellSoft's software:

------------------------------------------------------------------------------
This description is intended to assist with the use of HellSoft's 
bootpd.nlm.  You'll also need HellSoft's resolv.nlm.  You can find this
software via ftp at various sites, including the author's and mine;

novell.felk.cvut.cz - in nw311/bootpb and nw311/resolv
blaze.trentu.ca     - in pub/netware

All credit is due to HellSoft - I just use the stuff :-)

------------------------------------------------------------------------------
Here's a portion of our net;

                       CSD1.TRENTU.CA
                       192.75.12.139
                            +
                            +
=========================================================================
backbone                      +
                              +
                       192.75.12.138
                       CSD2.TRENTU.CA
                       +            +
                       +            +
               192.197.151.33   192.197.151.65

Off the backbone are several devices using bootpd.  The NW server CSD1 will 
*not* bootpd serve specific hosts (eg: x-terminals).  No hosts are served 
via bootpd from the backbone interface on CSD2 (192.75.12.138).  NW server CSD2 
will server any host on 2 interfaces (192.197.151.33 and 192.197.151.65).  No 
nameserver or gateway is defined for CSD2 bootpd clients.

Here are the essentials of the config files...

=======================
CSD1 - bootpd.cfg 
=======================
DisableFile	  sys:system/bootpd/disabled.cfg
Board   	  192.75.12.139
TimeToLive    1                      
Gateway       192.75.12.224
DNSServer     192.75.12.103
DNSServer     192.75.12.200
DynamicMask   0.0.0.225
DynamicName   CSD1-%d
DynamicHosts  30 192.75.12.170

=======================
CSD1 - diabled.cfg
=======================
Ethernet	00:00:A7:11:1C:33
Ethernet	00:00:A7:00:6E:B7
Ethernet	00:00:A7:11:1A:46
Ethernet	00:00:A7:11:1A:B2
Ethernet	00:00:A7:12:06:FF
Ethernet	00:00:A7:12:5A:3C
Ethernet	00:00:A7:13:6E:1C
Ethernet	00:00:A7:12:85:43
Ethernet	00:00:A7:12:0A:E3

=======================
CSD2 - bootpd.cfg
=======================
Board         192.197.151.33
TimeToLive    1                      
DynamicHosts  29 192.197.151.34
Board         192.197.151.65
TimeToLive    1                      
DynamicHosts  29 192.197.151.66
------------------------------------------------------------------------------

-- 
Ken Brown                                   internet: kbrown@trentu.ca
Trent University Computing & Telecommunications  tel: (705)748-1540
Peterborough, Ontario, Canada, K9J 7B8           fax: (705)748-1635

-----------[000353][next][prev][last][first]----------------------------------------------------
Date:      20 May 1994 12:20:17 +0100
From:      js@calvin.lif.icnet.uk ( BIU)
To:        comp.protocols.tcp-ip
Subject:   How to affect the connect timeout

This is probably a FAQ (but I couldn't find a FAQ, and
no-one pointed to where one was.)

When you try to connect a tcp socket using connect(2),
if the socket at the other end is listening but does
not accept the connection, the connect(2) call eventually
times out (at least this is my understanding).

Is there any way of setting the timeout time, or do you
have to use a timer to break into the call?
-- 
       Jack               js@biu.icnet.uk
                         
If you only have a hammer, you tend to see every problem as a nail.
                -- Maslow

-----------[000354][next][prev][last][first]----------------------------------------------------
Date:      Fri, 20 May 1994 14:22:14 GMT
From:      rstevens@noao.edu (W. Richard Stevens)
To:        comp.protocols.tcp-ip
Subject:   Re: How does a server detect a client crash using sockets?

> This I understand.  My question is what happens to things in buffers
> when a process goes down?

Discarded.

> Certainly the write() call will succeed even
> if a read() has not yet occurred at the other end.  Is there a way to know
> that the process at the other end of a socket connection has read the
> data that TCP has sent?

No.  You need an application-level ACK for this.  lpd uses this techinque.
That's why the use of SO_LINGER to force a close() to block until all
queued data is sent and ack'ed by the other "end", only tells you that the
other TCP has received and ack'ed the data.  It doesn't tell you that the
other application has read the data.

> Is there any way, within TCP, to know that the
> process at the other end of a socket has crashed and left unread stuff
> in a buffer?

Not that I'm aware of.  You can tell that the other end has crashed (see
my previous posting), but you have no information as to how much data
is discarded on both ends when this happens.  (I can imagine some groveling
through some kernel data structures to find out some information; for example
your send buffer sizes give you some information as to what's been sent;
TCP's sequence numbers tell you what's been sent and ACK'ed; the other
side's window advertisements tell you how much data the application has
read; etc.) but that seems pretty extreme.

	Rich Stevens

-----------[000355][next][prev][last][first]----------------------------------------------------
Date:      20 May 94 14:54:40 GMT
From:      hart@apanix.apana.org.au (Leigh Hart)
To:        comp.protocols.tcp-ip
Subject:   Re: Resale of registered

john.will@dscmail.com (John Will) writes:

>D >*Domain names*, on the other hand, are likely to be worth money,
>D >because they aren't interchangable.
 
>Why would a domain name be worth anything?  I can register as many 
>domain names as I can invent, doesn't seem to be a commodity...

You can't register the same one as an existing one.

A silly example would be if the current fad was to be groovy:

groovy.com.au

If you registered this domain name, I couldn't subsequently
register the same name.  If I *absolutely* _HAD_ to have it,
I would probably offer you money to transfer it's "ownership"
to me.

Having something that someone else wants makes it potentially
valuable.  having something that everyone wants is a commodity :-)

Cheers

Leigh
--
                                 Leigh Hart
                               C/- PO Box 758
                          North Adelaide  SA  5006
 hart@eppie.apana.org.au  hart@apanix.apana.org.au  hart@cleese.apana.org.au

-----------[000356][next][prev][last][first]----------------------------------------------------
Date:      20 May 1994 15:32:42 GMT
From:      jk@cis.ufl.edu (Jagadeesh Kasaraneni)
To:        comp.protocols.tcp-ip
Subject:   HELP NEEDED: Setting timeout on sockets?



Hello Everyone..

I already posted this once. I am posting this, again, for two reasons. There is 
a small correction to be made and I wanted to explain the application in a little
more detail.

I have a server and a client connected through sockets using TCP/IP. Operating 
System : DEC OSF v2.0.

In my application, server acts as a queue and clients are enqueuers and dequeuers.
When the queue is full it(server) doesn't process enqueuers' requests (by not
checking the enqueuers' connections for messages) and the 
enqueuers get stuck in trying to write the element to the server(queue). When
the enqueuer is very fast compared to the dequeuer it often gets blocked in the
write system call. Probably because the queue(server) ran out of buffer space
for the socket messages. At times, write gets blocked up to 40 seconds. After
some processing, the enqueuer crashes with a broken pipe signal. I caught the
signal and printed the errno and noticed that it is not set in anyway. I also
noticed the enqueuer has already waited for about 48 seconds, in write system
call, before getting the signal.

Well, here are my conclusions: Client was able to write as long as the buffer
space on the server side was available. Once it was full, client gets stuck in
the writing. When the time it was stuck was more than some predefined interval, 
system treated it as the broken connection and sent a SIGPIPE signal to client 
process.

Now, I would like to increase the timeout of this soket on the client side so that
it won't crash.  I read about SO_SNDTIMEO option for the sockets. But, according 
to Unix Network Programming by Richard  Stevens, this option right now has no 
effect on the timeouts used by TCP. 

Can anyone, out there, tell me whether my conclusions are correct and give me any 
suggestions about how to solve the problem? When can a process get a SIGPIPE
signal?

I, desperately, need a solution for this problem. Without the solution I can not
proceed with my work.

	Thanks a lot for any suggestion.
		Jagadeesh .K
		jk@cis.ufl.edu


-----------[000357][next][prev][last][first]----------------------------------------------------
Date:      Fri, 20 May 1994 17:10:12 GMT
From:      rstevens@noao.edu (W. Richard Stevens)
To:        comp.protocols.tcp-ip
Subject:   Re: HELP NEEDED: Setting timeout on sockets?

> I read about SO_SNDTIMEO option for the sockets. But, according 
> to Unix Network Programming by Richard  Stevens, this option right now has no 
> effect on the timeouts used by TCP. 

That was vanilla 4.3BSD.  It depends on the system nowadays.  4.3BSD Reno
and above support it, including AIX 3.2.x.  I have no idea whether OSF
supports it, however.

> When can a process get a SIGPIPE signal?

You should only receive SIGPIPE for a TCP socket when you write to the
socket and the socket has received an RST from the other end.  If you
have a read pending from the socket when it receives the RST, read()
returns an error and errno is set to ECONNRESET.

	Rich Stevens

-----------[000358][next][prev][last][first]----------------------------------------------------
Date:      Fri, 20 May 1994 17:21:19 GMT
From:      donp@novell.com (don provan)
To:        comp.protocols.tcp-ip
Subject:   Re: Resale of registered

In article <2rd5m3$qib@news.iastate.edu> john@iastate.edu (John Hascall) writes:
>make_money_fast()
>{
>	foreach company in fortune(500) do
>              if (!whois(company+".com")) register(company+".com")
>        done
>}

Just in case anyone takes this joke seriously: there is a quite
strenuous justification process for domain names. In particular, for
any given company name, the domain authorities would never consider
granting company-name.com to anyone other than that company. In fact,
it's fairly unlikely you could get *any* .com domain name assigned
except the one that reflected the name of *your* company. -don

-----------[000359][next][prev][last][first]----------------------------------------------------
Date:      Fri, 20 May 1994 17:30:00 GMT
From:      ihsan@nmti.com (jaleel ihsan)
To:        comp.protocols.tcp-ip
Subject:   Socket read question

On a Posix compliant system, if I make a blocking socket read request for N
bytes, is the system required to block until at least N bytes are received,
(unless a PUSH/OOB flag is set) or can the system unblock me for other reasons
(with < N bytes received) ?  I am writing the "other" side as well and can
ensure that N bytes are always sent and the PUSH/OOB flag is not set.  This
way I dont have to worry about < N bytes returned from the read request.

Thanks for you help.

-----------[000360][next][prev][last][first]----------------------------------------------------
Date:      Fri, 20 May 1994 18:24:19 GMT
From:      renaud@CERCA.UMontreal.CA (Alain Renaud)
To:        comp.protocols.tcp-ip
Subject:   Re: Source for WU ftpd available?

jtiller@ringer.jpl.nasa.gov (Jason Tiller) writes:

>Hello, all,
 
>Is the source for WU ftpd (4.2 now?) available freely?  If so, where
>would I find it?  If not, is another, similarly capable ftpd source
>available for a porting project?
 
>Thanks.

If you are taking about Wuarchive ftpd it's not at version 4.2 but 2.4
and you can get it at the ftp site "wuarchive.wustl.edu"

-- 
--------------- CEntre de Recherche en Calcul Applique ---------------
 Alain Renaud                          CERCA
 Senior system administrator           5160 Decarie Blvd. suite 400  
 Tel   : (514)369-5221                 Montreal (Quebec) Canada
 Fax   : (514)369-3880                 H3X 2H9
 Email : renaud@cerca.umontreal.ca

-----------[000361][next][prev][last][first]----------------------------------------------------
Date:      Fri, 20 May 1994 19:09:35 GMT
From:      jas@talking.COM (Jim Shankland)
To:        comp.protocols.tcp-ip
Subject:   Re: CTRL-ALT-DEL problem with sockets betn. PC and Unix

In article <2rh0aj$fr1@pyrnova.mis.pyramid.com> lstowell@pyrnova.mis.pyramid.com (Lon Stowell) writes:
>    Unfortunately there are now socket partners who don't send RST
>    on a socket before dying.
>    [...]
>    Just look for subjects with "Keepalives--were they really meant
>    for DOS PC's and their braindead operating systems and
>    intellectually challenged operators" or just the word "keepalive" 
>    in the header.

This reminds me a little of those humorous (or at least humorously
intended) signs one sees in offices announcing that "it has come
to management's attention that employees who are dying on the job
are failing to fall down."

It is impossible always to reset a TCP connection before dying,
simply because dying can come as a surprise (the system crashes,
a golden retriever trips over the power cord, a backhoe goes
through the fiber link, and other natural and man-made disasters).
And though one might reasonably include DOS in a list of man-made
disasters, it seems more a feature than a bug in DOS that the
system can be rebooted quickly and painlessly.  I think you're
too harsh on users who have come to rely on this feature.  Applications
for which this is a problem must resort to application-level or
socket-level keepalives, as you noted.

jas

-----------[000362][next][prev][last][first]----------------------------------------------------
Date:      Fri, 20 May 1994 19:27:36 GMT
From:      jas@talking.COM (Jim Shankland)
To:        comp.protocols.tcp-ip
Subject:   Re: How does a server detect a client crash using sockets?

In article <Cq2uzI.KJs@sgi1.fnet.cs.mci.com>,
abell@velveeta.apdev.cs.mci.com (Andrew_Bell) writes:

>This I understand.  My question is what happens to things in buffers
>when a process goes down?  Certainly the write() call will succeed even
>if a read() has not yet occurred at the other end.  Is there a way to know
>that the process at the other end of a socket connection has read the
>data that TCP has sent?  Is there any way, within TCP, to know that the
>process at the other end of a socket has crashed and left unread stuff
>in a buffer?  How can the sending process know what stuff never got read
>and needs to be retransmitted when a new connection is made?  Does this
>have to be done within the application (client and server)?  Does any
>UNIX OS with asynchronous I/O help the situation (write now and tell me
>when the read happened later)?

It has to be done at the application level, and even that doesn't
get you all the way there.  TCP can (eventually) tell you that the
connection has gone down, but has no way of knowing how much the
application has read or processed.  This is not a TCP (mis-)feature,
but is fundamental.  Consider the following dialogue between a teller
machine client and a server process at bank headquarters:

TM to Server:  Customer account xxx, password yyy,
	request $100 cash withdrawal.
Server to TM:  Request is valid, open cash drawer and dispense $100.
Network layer to server:  Circuit has gone down.

Now, has the cash been dispensed or not?  *The server can't tell.*
Even if the network layer somehow certifies that the TM's read()
call returned with the data the server sent, the TM might have
crashed immediately after the read() call returned, but before it
had done anything with the data -- the effect is the same as if
the data had never been delivered.

If the TM has a local disk, it can write to disk whether it has opened
the cash drawer or not.  Then, when the connection (or the TM) comes
back up, the server can ask the TM whether the last request before
the crash was processed.  Of course, that just shrinks the window (and
gets the network off the hook):  the TM could write to disk, "I opened
the cash drawer", and then crash before opening it; or it could open
the cash drawer and then crash before recording it on disk ....

jas

-----------[000363][next][prev][last][first]----------------------------------------------------
Date:      Fri, 20 May 1994 20:56:37 GMT
From:      jon@jonpc.po.com (Jonathan Lynn)
To:        comp.protocols.tcp-ip
Subject:   SLIP for Solaris 2.3

Anyone know if there is a public domain version of slip for Solaris 2.3?
Thanks

Jonathan Lynn


-----------[000364][next][prev][last][first]----------------------------------------------------
Date:      Sat, 21 May 94 09:36:00 -0600
From:      mark.stapleton@cld9.com (Mark Stapleton)
To:        comp.protocols.tcp-ip
Subject:   Re: SUMMARY: FTP problem

SB>"C'est la nuit qu'il est beau de croire a la lumiere." E. Rostand

Being a great fan of <<Cyrano deBergerac>>, I have to know if this 
Rostand quote is from that play. What is the English translation?

Mark

---
þ CmpQwk #UNREGþ UNREGISTERED EVALUATION COPY

-----------[000365][next][prev][last][first]----------------------------------------------------
Date:      Sat, 21 May 1994 00:59:54 GMT
From:      rleary@UMASSD.EDU
To:        comp.protocols.tcp-ip
Subject:   Half of IANA Ethernet Address Block?


  The MAC address block allocated to the IANA allows for 2**24 
individual addresses and 2**24 group addresses. Why (RFC1340) did 
the IANA halve this 2**24 group address space into 2**23 group 
addresses for Internet Multicast and 2**23 for "Assigned by IANA 
for other uses"? What are these "other uses"? I note that RFC1340 - 
Assigned Numbers - uses a number of addresses from the Internet Multicast 
half of the group space for specific assignment (e.g., 224.0.1.1 for 
NTP, and the range 224.0.3.0-224.0.3.255 for RFE Generic Service). 
(By "addresses from" I mean "IP multicast addresses that will map into 
the Internet Multicast half of the IANA MAC group address space".)

R. Leary
rleary@umassd.edu


-----------[000366][next][prev][last][first]----------------------------------------------------
Date:      Sat, 21 May 94 03:07:46 GMT
From:      ftlofaro@unlv.edu (Frank Lofaro)
To:        comp.protocols.tcp-ip
Subject:   Re: Resale of registered

In article <1994May20.172119.11988@novell.com> donp@novell.com (don provan) writes:
>In article <2rd5m3$qib@news.iastate.edu> john@iastate.edu (John Hascall) writes:
>>make_money_fast()
>>{
>>	foreach company in fortune(500) do
>>              if (!whois(company+".com")) register(company+".com")
>>        done
>>}
>
>Just in case anyone takes this joke seriously: there is a quite
>strenuous justification process for domain names. In particular, for
>any given company name, the domain authorities would never consider
>granting company-name.com to anyone other than that company. In fact,
>it's fairly unlikely you could get *any* .com domain name assigned
>except the one that reflected the name of *your* company. -don

Then explain this! :)

abyss.com is a domain name that isn't even for a company as far as I
can tell, just for an individual (who is into occult stuff and posts
to alt.magick, so I guess he found abyss.com to be a fitting name
:). He also in the process has himself registered in the whois
database. Pretty neat, eh? I think I want a cool domain name too! :)
In any event, here is the whois record:

6: 20:04:20: ~: reed$ whois "*abyss.com"
Thyagi NagaSiva Morgoth (ABYSS-DOM)
   871 Ironwood Drive
   San Jose, CA 95125

   Domain Name: ABYSS.COM

   Administrative Contact, Technical Contact, Zone Contact:
      NagaSiva, Thyagi  (TN86)  thyagi@CUP.PORTAL.COM
      (408) 266-2819

   Record last updated on 04-Jan-93.

   Domain servers in listed order:

   NOVA.UNIX.PORTAL.COM         156.151.1.101
   UUCP-GW-1.PA.DEC.COM         16.1.0.18
   UUCP-GW-2.PA.DEC.COM         16.1.0.19

   No known hosts under this secondary domain.

The InterNIC Registration Services Host ONLY contains Internet Information
(Networks, ASN's, Domains, and POC's).
Please use the whois server at nic.ddn.mil for MILNET Information.



-----------[000367][next][prev][last][first]----------------------------------------------------
Date:      21 May 1994 03:38:57 GMT
From:      olson@anchor.esd.sgi.com (Dave Olson)
To:        comp.sys.sgi.misc,comp.protocols.tcp-ip
Subject:   Re: [Q] SLIP problems

bill@msi.com (Bill Poitras) writes:

| I am connecting from a 486 computer with the Trumpet Winsock (a shareware 
| TCP/IP kernel for Windows)  over SLIP to an Silicon Graphics Workstation 
| running IRIX 4.0.5.  I can make the connection and telnet and FTP from 
| the PC to the SGI I connect to.  However I can't connect outside of the 
| machine to the rest of the network.  Here is the slip line the 
| /usr/etc/remoteslip script I am using:
| 
| exec /usr/etc/slip -i -r $USER
| 
| I don't think I can install any hard coded routes on the SGI I am logging 
| into, because it won't be the same one every time.  Basically I am 
| logging into a terminal server, and then telnetting to my SGI of choice 
| and logging in with a username identical to the hostname of my sliphost.  
| It may sound weird, but it works great.
| 
| Will changing the slip line like this in the /usr/etc/remoteslip script 
| help?
| 
| exec (route add host $USER localhost 0 ; \
|       /usr/etc/slip -i -r $USER ; \
|       route delete host $USER localhost)

No, you need to do the equivalent on the PC side.  The slip code already
installs a route to your PC on the server.  You can verify this with
rtquery or netstat -rn.

There *are* problems with Cisco (and doubtless other routers) that
can't understand host routes, particularly when routes to the same
base net go through different slip servers.  Very recent cisco
firmware is supposed to be better about this.

Now, if you aren't running routed on the sgi server to advertise
the route, you'll have to do something about that (or use gated,
or use static routes *on the router to the net that your slip
server is on, not on the slip server*.  But static routes have problems
of their own).
--

The most beautiful things in the world are              |   Dave Olson
those from which all excess weight has been             |   Silicon Graphics
removed.  -Henry Ford                                   |   olson@sgi.com

-----------[000368][next][prev][last][first]----------------------------------------------------
Date:      Sat, 21 May 1994 04:56:36 GMT
From:      chang@cp1.bell-atl.com (chang)
To:        comp.protocols.tcp-ip
Subject:   TCP sessions limitation on UNIX V.4?

	We have a network that consists of an StarServer E (3 Intel CPUs) and
around 200 X-terminals. Recently we have this strange problem: when the server
has about 1024 TCP sessions (including both ESTABLISHED and CLOSED), users
will periodically losing their X-window sessions. The kernel was configured
for 2048 tcp sessions, but the problem would not go away. AT&T worked on this
problem for about two weeks and concluded the tcp driver can only support
1024 sessions. Is this true? I don't have source code, but 1024 seems a
little too low for me.

	My question is, when I bump up the max tcp minor number in the
'/etc/conf/cf.d/mdevice' file to increase the tcp number, do I need to 
bump up something else too? Most of the TCP sessions on my system were made
by the 'rlogin'. Anyone has the similar problem as I have? If you have more
than 1024 tcp sessions on your Intel x86 V.4 system, can you kindly mail me 
your '/etc/conf/cf.d/mdevice', '/etc/conf/sdevice/tcp' and 
'/etc/conf/sdevice/ptm' files?

	I'm not sure when is AT&T going to give me the new driver. At the
mean time, any help or suggestions will be deeply appreciated.

Thanks,
chang

e-mail: chang@cp1.bell-atl.com
Phone:  (301) 236-2619 (W)
	(301) 236-8598 (Fax)

-----------[000369][next][prev][last][first]----------------------------------------------------
Date:      21 May 1994 06:14:49 GMT
From:      antonyc@linuxftp.caltech.edu (antony)
To:        comp.sys.sun.admin,comp.protocols.tcp-ip,comp.dcom.lans.ethernet
Subject:   Re: Two IP networks on one ethernet = broadcast storms

>Jeff Makey (makey@VisiCom.COM) wrote:
 
>As an aside, if you have two interfaces onto the same cable, check
>that they have different MAC addresses. As root, ifconfig them to 
>check this out. If these are the same, it would be a good idea to 
>change one of them; if one starts 8:0:20 then make the other 9:0:20...
>The reason for this is that any ethernet frame sent to the 8:0:20...
>mac address will be received on both interfaces. Luckily the IP
>layer can handle duplicate datagrams :-)

i think he meant 10:0:20 ; this changes the u/l bit instead of the
multicast bit.

-----------[000370][next][prev][last][first]----------------------------------------------------
Date:      21 May 1994 09:15:56 GMT
From:      tli@cisco.com (Tony Li)
To:        comp.protocols.tcp-ip
Subject:   Re: Assigning IP addresses to a CISCO WAN

In article <1994May19.131352.19239@cyantic.com> mark@cyantic.com (Mark T.
Dornfeld) writes: 
    
    Each LAN has its own class C IP subnet, and currently the two synchronous
    ports comprise a third class C subnet. 
    
    LAN1     ---------------------------------     192.246.150.x
                           |
                        --------
                        |  E1  |
               CISCO1   |  S1  |    192.246.151.1
                        --------
                           |
          56K SYNC WAN     |
                           |
                        --------
                        |  S2  |    192.246.151.2
               CISCO2   |  E2  |
                        --------
                           |
    LAN2     ---------------------------------     192.246.152.x
    
    Is this the correct way to number the WAN?  

"correct" is in the eye of the beholder.  ;-)

Certainly what you specify is techically acceptable.  However, as you also
note, it's very wasteful of IP address space.  At the very least, you might
use a subnet of one of your other networks.

    Should the WAN have it's own
    subnet, or should interface S1 have an address from LAN2 and interface S2
    have a number from LAN1.  

Either, as it's up to you.  You should look at the descripton of "ip
unnumbered" in the manual.  It discusses the tradeoffs there...

Tony

-----------[000371][next][prev][last][first]----------------------------------------------------
Date:      21 May 1994 10:37:19 GMT
From:      jenkins@oils (Jon Jenkins)
To:        comp.protocols.tcp-ip
Subject:   Re: Resale of registered

BIU (js@calvin.lif.icnet.uk) wrote:
: In article <2rd5m3$qib@news.iastate.edu> john@iastate.edu (John Hascall) writes:
: >John Will <john.will@dscmail.com> wrote:
: >}Why would a domain name be worth anything?  I can register as many 
: >}domain names as I can invent, doesn't seem to be a commodity...
: >
: >make_money_fast()
: >{
: >	foreach company in fortune(500) do
: >              if (!whois(company+".com")) 
		    {
		    register(company+".com")
		    inform_system_manager(company+".com");
		    }
: >        done
: >}

-----------[000372][next][prev][last][first]----------------------------------------------------
Date:      Sat, 21 May 1994 11:02:09 +0000
From:      proyse@peeras.demon.co.uk (Phil Royse)
To:        comp.protocols.tcp-ip
Subject:   Re: Resale of registered

In article <hart.769445680@apanix.apana.org.au>
           hart@apanix.apana.org.au "Leigh Hart" writes:

>john.will@dscmail.com (John Will) writes:
>
>>D >*Domain names*, on the other hand, are likely to be worth money,
>>D >because they aren't interchangable.
>
>If you registered this domain name, I couldn't subsequently
>register the same name.  If I *absolutely* _HAD_ to have it,
>I would probably offer you money to transfer it's "ownership"
>to me.
>
>Having something that someone else wants makes it potentially
>valuable.  having something that everyone wants is a commodity :-)
>
>Cheers
>
>Leigh

Yeah.

Which brings us round to my original question, which was not
about domain names, but IP address spaces....

Another way of phrasing the question:

	"Will IP address space *exhaustion* on the Internet be
	heralded by the arrival of advertisements "Small Co.
	in UK seeks IP address space for 2055 hosts.  Must be 
	contiguous for CIDR.  Offers less than US$5000
	considered."....?"

Am I being  too naiive or too commercial?

Phil

-- 

Phil Royse     Comms Consultant  |  PRA Consulting Ltd.
TUDOR HOUSE                      |
12 Woodside Road, Purley
Surrey  CR8 4LN (UK)  Tel: (+44) 81 645-9868   proyse@peeras.demon.co.uk

-----------[000373][next][prev][last][first]----------------------------------------------------
Date:      21 May 94 11:09:39 GMT
From:      avalon@cairo.anu.edu.au (Darren Reed)
To:        comp.protocols.tcp-ip
Subject:   Turning off TCP/IP options


A couple of queries I have about TCP/IP options with TCP connections...

If I open a connection with various options set, and the other end strips
off the options (like what tcpd does), will my end continue to use them
whilst the other end doesn't, will both drop or both continue to do so ?

If the options keep being presented to the remote end, will it try and
"learn" about them and put them back after they're taken away or are
they only given to the socket when the connection establishes and then
must either be manually be added/taken away with set/getsockopt ?

Are there any TCP options which can be considered dangerous or is it
just IP options (source routing in particular) that we need to be wary
of ?

Thanks,
Darren

-----------[000374][next][prev][last][first]----------------------------------------------------
Date:      Sat, 21 May 1994 11:31:16 GMT
From:      arshad@con-sys.demon.co.uk (Arshad Mahmood)
To:        comp.protocols.tcp-ip
Subject:   Realtime database

Hi,
	We would be interested to know if there are any products
out there which would allow clients across a heterogenous 
TCP/IP network to share data in a manner which AI specialists might
call a blackboard. For those not familiar with the term "blackboard",
you can imagine it simply as an array where each element may be some
piece of data, essentially some individuals produce data which goes
into one of these slots and others are interested in it (and are told
when it has been produced, this is grossly oversimplifed but you get 
the idea). 

   A primary concern is speed since some of this
data will arrive in real-time. The second is heterogneity (spelling?),
since our main applications must run on IBM RS6000, SUN and HP (series 700), 
the realtime
data being collected by a VME bus based system running OS-9. So this
application must be able to run on each of these systems, further
it must be in the form of a library which all our applications can 
make use of.

  We built something like this for an earlier project, however it would take
some 6 months to bring it up to scratch for any real use.

  If you know of such a system please send a reply by e-mail.

Regards,
A. Mahmood
Senior Software Engineer
Concept Systems Ltd
Edinburgh
e-mail: arshad@con-sys.demon.co.uk


-- 
-----------------------------

-----------[000375][next][prev][last][first]----------------------------------------------------
Date:      Sat, 21 May 1994 14:10:07 GMT
From:      fenner@cmf.nrl.navy.mil (William C. Fenner)
To:        comp.protocols.tcp-ip
Subject:   Resale of registered

In article <1994May20.172119.11988@novell.com>,
don provan <donp@novell.com> wrote:
>Just in case anyone takes this joke seriously: there is a quite
>strenuous justification process for domain names. In particular, for
>any given company name, the domain authorities would never consider
>granting company-name.com to anyone other than that company. In fact,
>it's fairly unlikely you could get *any* .com domain name assigned
>except the one that reflected the name of *your* company. -don

Yeah?  How did SprintLink manage to get the MCI.NET domain, then?  (MCI
got the NIC to take it back in mid-April, but...)

In addition, there are quite a number of businesses with the same
initials.  If IBM.COM wasn't already registered, what's to stop me from
registering my business called Internal Banana Makers under IBM.COM?

  Bill
-- 
Bill Fenner                  fenner@cmf.nrl.navy.mil

-----------[000376][next][prev][last][first]----------------------------------------------------
Date:      21 May 1994 13:21:00 +0100
From:      kwbes@kwbes.sh.sub.de (Karl-Werner Beszus)
To:        comp.protocols.tcp-ip
Subject:   Re: Mac to PC TCP-IP link

In article <2r56qr$lhf@paperboy.osf.org>,
dswartz@pugsley.osf.org (Dan Swartzendruber)  wrote:

> In article <dafordCpoBn2.2w8@netcom.com> daford@netcom.com
> >(David Ford) writes:
> >A friend of mine has a Mac and PC and wants to connect them
> >together without going apple-talk like the Coactive

[stuff deleted]

> thing to be wary of is the Mac ethernet interface.  Any Mac
> which claims to have "builtin ethernet" doesn't really.  At
> least not with a compatible interface.  They use some special
> Apple-only socket and electronics.  You need a gadget which
> plugs into the Mac port and provides 10Base2 (or whatever) on
> the other.  The ripoff is that this is almost as expensive as

[stuff deleted]

Thats nearly true, if you want a plug 'n play solution.

If you can afford it to solder new connections to the builtin  
interface in the Mac, then you can use this interface directly.

I saw once a description how to do it in a magazin.
--
+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
        Karl-Werner Beszus, Eichenweg 5, 25584 Besdorf
          voice, fax, etc.: +49-4827-2637  Germany
                     kwbes@kwbes.sh.sub.de
+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++

-----------[000377][next][prev][last][first]----------------------------------------------------
Date:      Sat, 21 May 1994 15:31:31 GMT
From:      vjs@calcite.rhyolite.com (Vernon Schryver)
To:        comp.protocols.tcp-ip
Subject:   Re: Resale of registered

In article <1994May20.172119.11988@novell.com> donp@novell.com (don provan) writes:
>In article <2rd5m3$qib@news.iastate.edu> john@iastate.edu (John Hascall) writes:
>>make_money_fast()
>>{
>>	foreach company in fortune(500) do
>>              if (!whois(company+".com")) register(company+".com")
>>        done
>>}
>
>Just in case anyone takes this joke seriously: there is a quite
>strenuous justification process for domain names. In particular, for
>any given company name, the domain authorities would never consider
>granting company-name.com to anyone other than that company. In fact,
>it's fairly unlikely you could get *any* .com domain name assigned
>except the one that reflected the name of *your* company. -don

Don Provin must also be joking.
Except for the fortune 10 and a few odd computer outfits, the NIC is
not likely to notice or care.  If Joe's Apple Farm asks for apple.com,
how is the NIC to know if it's the same Joe who runs Joe's General
Mutagens who asked for gm.com?  (Yes, I know apple.com was registered
a long time ago, and not by the music people.)

I personally registered two domains, sgi.com and rhyolite.com, and didn't
provide any documentation at all.  All I did was send in the electronic
form.  The only hassle was that they wouldn't let me be technical contact
for both of them.  They probably noticed only because I tried to use my
official handle for both, and surely would not have noticed 500 varitions
of Joe, Josef, Joseph, or Joan Smith or Jones.

Somehow I doubt that mean Internet Services Providers who have and
continue to register zillions of domains for their customers, starting
with Uunet years ago, need to provide much documentation to the NIC.


Vernon Schryver    vjs@rhyolite.com

-----------[000378][next][prev][last][first]----------------------------------------------------
Date:      Sat, 21 May 1994 19:15:31 GMT
From:      gnn@netcom.com (George Neville-Neil)
To:        comp.protocols.tcp-ip
Subject:   Re: UDP SOCKET HELP

vbraco@news.delphi.com (VBRACO@DELPHI.COM) writes:

>I'm trying to write a program that will send data out on a socket which
>can be read by one or more programs on other hosts. The hosts are HP's.
>The HP documentation has led me to believe that there is a way to
>'broadcast' this data in such a way that it can be received by multiple

This depends, to some extent, the nature of your network.  This should
definitely work if all of the hosts are on the same physical ethernet.

>hosts simultaneously. I have a pair of simple prototype programs which
>communicate using send() & recv() but the 'receiver' only seems to get the
>message when the 'sender' does a connect() before the send() or if
>sendto() is used. In each case the IP address of the host running the
>'receiver' must be explicitly specified or the 'receiver' doesn't see the
>message. I have tried specifying the wildcard address and using the
>setsockopt() function to set SO_BROADCAST but so far - no joy.  No
>response from the 'receiver' is desired, however, there may be several
>'receiver' processes on different hosts all wishing to read the 'sender's
>data.  Anybody know how to do this? A code fragment would be very much
>appreciated. If this is documented anywhere, please take a moment to point
>me at it.


You should take a look at the FAQ, which will be reposted on June 1.

In the meantime take a look at code in these places (from the FAQ):
A)  Code from the Internetworking with TCP/IP Volume III is available
for anonymous ftp from:

arthur.cs.purdue.edu:/pub/dls

Code used in the Net-2 version of Berkeley Unix is available for
anonymous ftp from:

ftp.uu.net:systems/unix/bsd-sources/sys/netinet

and

gatekeeper.dec.com:/pub/BSD/net2/sys/netinet

Code from Richard Steven's book is available on:
ftp.uu.net:/published/books/stevens.*



There should be at least one program in the Comer and Stevens stuff
that does this, probably more.

Hope this helps.

Later,
George

-- 
gnn@netcom.com

Gentelmen, I will not have you fighting in the War Room.
					--- The President in Dr. Stragelove

-----------[000379][next][prev][last][first]----------------------------------------------------
Date:      22 May 1994 03:49:55 GMT
From:      john@iastate.edu (John Hascall)
To:        comp.protocols.tcp-ip
Subject:   Re: Resale of registered

Vernon Schryver <vjs@calcite.rhyolite.com> wrote:
}Don Provin must also be joking.
}Except for the fortune 10 and a few odd computer outfits, the NIC is
}not likely to notice or care.  If Joe's Apple Farm asks for apple.com,
}how is the NIC to know if it's the same Joe who runs Joe's General
}Mutagens who asked for gm.com?  (Yes, I know apple.com was registered
}a long time ago, and not by the music people.)

I wonder how they verified this one...

pooh.cc> whois hell.org
hellnet (HELL-DOM)
   818 Burnett Ave
   Ames, IA 50010

   Domain Name: HELL.ORG
   ...


:)

John
-- 
John Hascall                   ``An ill-chosen word is the fool's messenger.''
Systems Software Engineer
Project Vincent
Iowa State University Computation Center  +  Ames, IA  50011  +  515/294-9551

-----------[000380][next][prev][last][first]----------------------------------------------------
Date:      22 May 1994 04:35:25 GMT
From:      barmar@think.com (Barry Margolin)
To:        comp.protocols.tcp-ip
Subject:   Re: Socket read question

In article <id.W3O91.ENB@nmti.com> ihsan@nmti.com (jaleel ihsan) writes:
>On a Posix compliant system, if I make a blocking socket read request for N
>bytes, is the system required to block until at least N bytes are received,

Read() on a socket is the same as read() on any other device.  The size
argument is just a maximum, not a requirement.  Stream sockets are byte
streams -- they don't include message boundaries.  Consider that if you
didn't know the size of the incoming message you would presumably provide a
huge buffer, which would be pretty troublesome if it blocked until that
amount were read.

-- 
Barry Margolin
System Manager, Thinking Machines Corp.

barmar@think.com          {uunet,harvard}!think!barmar

-----------[000381][next][prev][last][first]----------------------------------------------------
Date:      22 May 1994 12:44:12 GMT
From:      mjo@iao.ford.com (Mike O'Connor)
To:        comp.protocols.tcp-ip
Subject:   Re: Resale of registered

:: >make_money_fast()
:: >{
:: >	foreach company in fortune(500) do
:: >              if (!whois(company+".com")) 
:		    {
:		    register(company+".com")
:		    inform_system_manager(company+".com");
:		    }
:: >        done

Rather than seeing people trying to register hunks of the .com DNS
just to tick people off, I'd rather see someone work on the system
call that enables me to figure out who the true person in charge is
with some of these large organizations.  :)

-- 
 Michael J. O'Connor           |  Internet:  mjo@jobone.srl.ford.com
 Ford Motor Company, OPEO      |  UUCP:      ...!fmsrl7!opeo!mjo
 20000 Rotunda, Bldg. 1-3001   |  Phone:     +1 (313) 248-1260
 Dearborn, MI  48121           |  Fax:       +1 (313) 323-6277

-----------[000382][next][prev][last][first]----------------------------------------------------
Date:      Sun, 22 May 94 13:01:55 BST
From:      db15@ukc.ac.uk (Damiano Bolla)
To:        comp.protocols.tcp-ip
Subject:   Any Public Domain Simple router software available ?

I have to manage a simple network like this

	11	12	13
	|	|	|
	R1	R2	R3
	------------------- net 2 ------- cisco --- World

11,12,13,2 are subnets of a Class B network with netmask 255.255.255
The routers (R) are Unix machines and have a static routing table
to avoid problems at boot. (Some mounts directory from others)
(I know that it should work using in.routed, it just results in 
 a quite long routing table + some weird routing of the 
 default router )

The cisco had a static routing table but now the admin wants
to go dinamic I.e. routers R1,R2,R3 should tell to the cisco 
that they can route to the given subnet. 

The question is:
Can I keep a static routing table into router R1,R2,R3 and 
in the same time tell Cisco that they are routers for
the given subnets ? 

I found source code for in.routed and egp and I am willing to 
work on it to do what I want but I am wondering if this is 
the way to go or not. 
The machines are Sun, Solaris 2.3. But I may "convert" one
to a 4.1.3.

Damiano


-----------[000383][next][prev][last][first]----------------------------------------------------
Date:      Sun, 22 May 1994 18:47:15 GMT
From:      cliffb@skeeter.bdb.com (Cliff Bedore)
To:        comp.protocols.tcp-ip
Subject:   LCIS Saga Continues (Comer Books)

Just a short note to say I received my two missing volumes of the Comer set.
Don't know if they read this newsgroup or it was just luck. (Don't care either
:-) )  I now have no more excuses to be ignorant of the inner workings of
TCP/IP.  

Cliff

-----------[000384][next][prev][last][first]----------------------------------------------------
Date:      23 May 1994 02:15:53 GMT
From:      jk@cis.ufl.edu (Jagadeesh Kasaraneni)
To:        comp.protocols.tcp-ip
Subject:   Re: HELP NEEDED: Setting timeout on sockets?


|> When can a process get a SIGPIPE signal?
 
|You should only receive SIGPIPE for a TCP socket when you write to the
|socket and the socket has received an RST from the other end.  If you
|have a read pending from the socket when it receives the RST, read()
|returns an error and errno is set to ECONNRESET.
 
|	Rich Stevens


I was doing some experiments with the sockets to understand more about the
behaviour of the TCP sockets when one end is closed. The experimental system has
a server and a client.

The server waits for a connection from a client. As soon as a connection is 
established,it immediately closes the connection and quits.

The client establishes a connection with the server. It waits till the server
quits ( I made it wait, of course!) and for three times it attempts to read from
the server and three times to write something to server. These are my
observations:

	* Reads written with a return value 0. And the errno is set to "Error 0".
	  No signal is generated.
	
	* First write is successfull with a return value equal to the number of
	  bytes. errno is "Error 0". No signal generated.

	* Second write is unsuccessfull with a return value -1. errno is "broken
	   pipe". SIGPIPE is generated.
	* Same with the third write.

Now my questions are:

1. When is this RST sent?
2. Why didn't the read return an error with errno set to ECONNRESET?
3. Why was the first write successful?

	Thanks a lot for any answers..

		Jagadeesh
		jk@cis.ufl.edu



-----------[000385][next][prev][last][first]----------------------------------------------------
Date:      Mon, 23 May 1994 10:51:44 MST
From:      David.Bear@Asu.Edu
To:        comp.protocols.tcp-ip
Subject:   free socket libary for os2

Does anyone know of a free socket library for OS/2 tcpip?  
I would like to play around with sockets, but not for 500 bucks...

David Bear
Support Systems Analyst
College of Public Programs
internet: David.Bear@asu.edu
phone: 602-965-8257


-----------[000386][next][prev][last][first]----------------------------------------------------
Date:      Mon, 23 May 94 04:16:09 GMT
From:      hal9001@panix.com (Robert A. Rosenberg)
To:        comp.protocols.tcp-ip
Subject:   Re: MacTCP & SLIP Help!!

In Article <2rnfjd$rg6@cleese.apana.org.au>, bridgway@cleese.apana.org.au
(Brian Ridgway) wrote:
>
>I am having problems setting up MacTCP and InterSLIP for dialup SLIP
>access. 
> 
>With these installed as per instructions, applications such as LeeMail,
>Newswatcher and Telnet all hang up immediately after launching.
> 

Check if your provider (the people who you are dialing into) provide PPP
support as well as SLIP support. If that do, ditch the InterSlip and use the
Freeware MacPPP. Not only is MacPPP a better program but a PPP connection is
more robust and has less overhead.

-----------[000387][next][prev][last][first]----------------------------------------------------
Date:      Mon, 23 May 94 05:03:43 GMT
From:      ftlofaro@unlv.edu (Frank Lofaro)
To:        comp.protocols.tcp-ip
Subject:   Re: MacTCP & SLIP Help!!

In article <hal9001.1120054209B@news.panix.com> hal9001@panix.com (Robert A. Rosenberg) writes:
>In Article <2rnfjd$rg6@cleese.apana.org.au>, bridgway@cleese.apana.org.au
>(Brian Ridgway) wrote:
>>
>>I am having problems setting up MacTCP and InterSLIP for dialup SLIP
>>access. 
>> 
>>With these installed as per instructions, applications such as LeeMail,
>>Newswatcher and Telnet all hang up immediately after launching.
>> 
>
>Check if your provider (the people who you are dialing into) provide PPP
>support as well as SLIP support. If that do, ditch the InterSlip and use the
>Freeware MacPPP. Not only is MacPPP a better program but a PPP connection is
>more robust and has less overhead.
                 ^^^^^^^^^^^^^^^^^^

Bwahahahaha!!! PPP has _less_ overhead! Get real! SLIP just uses one character 
to indicated end of packet, another (ESC) to allow escaping the END character, 
and itself if it occurs in a data stream, and 2 others that are used with the 
escape to indicate an escaped end or escaped escape character. 4 characters 
total, only 2 of which need to be worried about (escaped) if they appear in 
a data stream.

SLIP is nice, efficient and simple. PPP adds all sort of link-level cruft, 
and extra negotiation. With SLIP to just receive and send IP datagrams. With 
PPP you have to worry about opening the link level layer, sending a 
whole PPP packet header along with your IP datagram, etc. And then closing 
it down when you are all done. You have to worry about as many state 
transitions as you do for TCP (a heavyweight _transport_ level protocol; 
that amount of complexity is NOT needed for the link level by a long shot. 
Oh, there is MUCH more chance of a buggy PPP implementation, it takes a 
special breed of incompetant programmer to screw up SLIP.

Check out the SLIP (1055) and PPP (I forget the number of the week, they 
keep changing it) RFC's, and you'll see what I mean!

As for more robust, if the link is screwed, so are you! If the link takes a 
hit, but stays up, SLIP will just start working again (no need to "recover").
PPP might start a link close down sequence, which would then have to be 
aborted, etc. A link that is very intermittent might be unusuable under PPP 
but be passible under SLIP.

P.S. Sorry if the abrasiveness of my initial comments offend, but I almost 
fell back in my chair when I read this.



-----------[000388][next][prev][last][first]----------------------------------------------------
Date:      Mon, 23 May 1994 07:32:53 GMT
From:      rc@netarch.com (Ran-Chi Huang)
To:        comp.protocols.tcp-ip
Subject:   Help with BSD sockets

I am writing a client server application and it essentially works except
for one minor problem.

When I kill the daemon appsd via "kill -9 17897", the daemon indeed
goes away.

    root 17897   111 80 23:49:02 ?          0:03 appsd

However, when I try starting it up again, I cannot start it. Upon
checking I find the following lying around when I do a "netstat -a".
After I wait for 3-4 minutes, these TIME_WAIT connections disappear
and I am able to start the daemon again.

solaris.6096         solaris.42043        16384      0 16384      0 TIME_WAIT
solaris.6096         solaris.42045        16384      0 16384      0 TIME_WAIT
solaris.6096         solaris.42047        16384      0 16384      0 TIME_WAIT

How do I fix it up so that I can restart the daemon immediately when
I kill it ?

Please email me your responses and I will summarize.

Thanks !
-- 
--
Ran-Chi Huang                                               rc@netarch.com
Network Architecture Consulting., Fremont, CA                   netarch!rc

-----------[000389][next][prev][last][first]----------------------------------------------------
Date:      23 May 94 18:00:57 -0500
From:      harvey@indyvax.iupui.edu (James Harvey)
To:        comp.protocols.tcp-ip,comp.protocols.tcp-ip.domains
Subject:   Re: Weird DNS/Nameserver problems after changing IP address with NIC

In article <2rj1jk$b3v@hsc.usc.edu>, hvora@hsc.usc.edu (Heena Vora) writes:
> Hi all,
> Ignore the earlier post (it was incomplete)
>
> I'm helping some friends set up an internet node, and am encountering strange
> nameserver problems
>
> We registered with the NIC and obtained an IP address of 198.147.76.1.  When we
> actually connected through our service provider, this address  had to be subnetted
> because the provider was short of network numbers.  So we became 198.147.76.65 on
> network 198.147.76.64.
>
> We informed the NIC and our secondary to make the change, which they did.  All this was
> about 10 days ago.  Since then, I've had trouble telnetting to a couple of
> nodes (such as usc.edu) for a random period of about 10 minutes or more
> during the day.  I'll get a message saying that the TELNET host has to close
> the session because it can't determine our domain name from our IP address.

WHOIS at the NIC shows the class C address 198.147.76 is in a block
registered to Los Nettos:

% whois 198.147.76
Los Nettos (NETBLK-LS-NTTS-BLK198)
   Los Nettos
   USC/ISI
   4676 Admiralty Way Suite 1000
   Marina del Rey, CA 90292-6695

   Netname: NETBLK-LS-NTTS-BLK198
   Netblock: 198.147.64.0 - 198.147.127.0

   Coordinator:
      Prue, Walt  (WP8)  PRUE@ISI.EDU
      (310) 822-1511

   Record last updated on 01-Oct-93.

The InterNIC Registration Services Host ONLY contains Internet Information
(Networks, ASN's, Domains, and POC's).
Please use the whois server at nic.ddn.mil for MILNET Information.
%

Los Nettos has not registered for DNS service for 76.147.198.IN-ADDR.ARPA,
which is why there is no way for other sites to determine your domain from
your IP address.  You need to talk to your service provider about this.
Ideally if you had a whole class C net they could delegate the IN-ADDR.ARPA
domain for it to you (presumably hollywood.cinenet.net?), otherwise they
are going to have to maintain PTR RRs for all the hosts in your subnet.

I have set followups to comp.protocols.tcp-ip.domains.
--
James Harvey   harvey@iupui.edu   IUPUI OIT Networks and Systems Tech Support
Disclaimer:  These are my own opinions.  I do not speak for Indiana University.

-----------[000390][next][prev][last][first]----------------------------------------------------
Date:      23 May 94 15:04:56
From:      zhao@crl.nmsu.edu (Z. Zhao)
To:        comp.protocols.tcp-ip.ibmpc,comp.protocols.tcp-ip,comp.protocols.ppp
Subject:   security of SLIP/PPP

I want to connect some PCs in my company to a SPARC mail-server by
using SLIP/PPP. My boss asks me: "Is it secure?" I cannot answer. 

I assume there was no firewall installed on the mail-server. Is
SLIP/PPP connection are more secure, or there are easier solution on
security issues, comparing with ethernet connection?


ZiZi

-----------[000391][next][prev][last][first]----------------------------------------------------
Date:      Mon, 23 May 1994 10:20:50 +0000
From:      richard@corixia.demon.co.uk (Richard Ashton)
To:        comp.protocols.tcp-ip
Subject:   Re:   !!!

In article <ABPB8tjGO3@bfh.irkutsk.su> denis@bfh.irkutsk.su " denis" writes:

}             !
}
}      ,     NetWare 3.11  - EverEX Step-VL
}
}   P.S.    :  - Seagate ST3600N.
}                                                       

        Thats the prettiest for months and makes as much sense as
        some.


Regards   {} {} {}   Richard   {} {} {}   TEAM OS/2

-----------[000392][next][prev][last][first]----------------------------------------------------
Date:      Mon, 23 May 1994 10:47:19 GMT
From:      mw00378@lime.sbil.co.uk (Malcolm White)
To:        comp.protocols.tcp-ip
Subject:   Re: IGRP proprietary

In article ibi@ghost.sm.dsi.unimi.it, villa@ghost.sm.dsi.unimi.it (Max .Vilas. Villa) writes:

|Hi, I need to know if the IGRP protocol (CISCO's routing protocol used 
|in internet) is proprietary of CISCO or other manufacturers handle that.
|Please send answers (if any) to villa@dsi.unimi.it.
|Vilas
|-- 
|Massimo .Vilas. Villa (villa@dsi.unimi.it)
|System and News Administrator
|Computer Science Dep. - Milan University 

IGRP is propietary to Cisco and is not available from any other manufacturers. If you are using IGRP and wish to communicate with another vendors router you will need to redistribute IGRP into OSPF, for example.

Regards, Malcolm White
Salomon Brothers, London





-----------[000393][next][prev][last][first]----------------------------------------------------
Date:      Mon, 23 May 94 12:20:00 +0000
From:      jim.mcneill@almac.co.uk (Jim Mcneill)
To:        comp.protocols.tcp-ip
Subject:   Trumpet Winsock

Anyone tell me where I can get a hand on this Trumper Winsock thingie?

Jim
---
 * 1st 1.11 #324 * 

-----------[000394][next][prev][last][first]----------------------------------------------------
Date:      22 May 1994 20:57:23 +0930
From:      bridgway@cleese.apana.org.au (Brian Ridgway)
To:        comp.protocols.tcp-ip
Subject:   MacTCP & SLIP Help!!


I am having problems setting up MacTCP and InterSLIP for dialup SLIP
access. 
 
With these installed as per instructions, applications such as LeeMail,
Newswatcher and Telnet all hang up immediately after launching.
 
I have tested for system extension conflicts and even reinstalled system
software but to no avail.
 
Relevant software:
System 7.1
MacTCP 2.0.2
InterSlip 1.0fc3
Leemail 2.0.3
NewsWatcher 2.0d
Telnet 2.5
Hardware:
LC ll 6 MB RAM
 
Can anyone help?
 
 
 


-----------[000395][next][prev][last][first]----------------------------------------------------
Date:      23 May 1994 12:50:09 -0000
From:      vbraco@news.delphi.com (VBRACO@DELPHI.COM)
To:        comp.protocols.tcp-ip
Subject:   Re: UDP SOCKET HELP

Thanks for the pointers!

BTW - Love that sig from Dr. Starngelove...

Vince


-----------[000396][next][prev][last][first]----------------------------------------------------
Date:      23 May 1994 13:36:47 GMT
From:      P.E.Atkins@SUNS.salford.ac.uk (Peter Atkins)
To:        comp.protocols.tcp-ip
Subject:   Novell Conference




             N O V E L L    C O N F E R E N C E

                    I N V I T A T I O N



Salford University Network Services (SUNS) is once again hosting the
Novell Technical Conference, to be held at the University between the
11th and the 14th of July this year. Following the trend set by last years
conference we will be running several parallel sessions during the four
days and have an exhibition made up of network related computer
suppliers. The majority of the presentations will be given by members of
Novells' European Enterprise Solutions Division and senior Novell staff
from Novell's headquarters in Utah will be present throughout the event.
The new Novell Vice President in charge of Education, Dan Bagwell, will
also be present throughout the conference and this is our opportunity to
educate him concerning the different needs of the UK community compared
with the US market. Both corporate and educational technical staff will 
be invited to the conference and we hope to attract approximately 500 people. 

The agenda has not yet been finalised but the topic areas to be covered
are outlined below. The content of the sessions will be of a technical
nature and sales talk will be restricted to the exhibition. Time will be
scheduled to allow you to discuss your particular interests/problems
with Novell staff and other conference attendees. 

AS AN ADDED BONUS NOVELL HAVE AGREED TO GIVE A COPY OF NOVELL APPWARE TO
ALL DELEGATES ATTENDING THIS YEARS CONFERENCE.

In the hope that you will be able to join us I have attached a conference 
registration form which should be returned to myself as soon as possible. 
As the conference is of a technical/training nature the fees will not be 
subject to VAT.



Peter Atkins

-----------------------------------------------------------------------

                 Novell Technical Conference 1994

                          Outline Agenda


The following topics will be covered during the conference:

    AppWare Foundation
    Directory Services' Objects and Rights
    Distributed Network Analysis Using the NetWare LANalyzer Agent
    How AppWare fits with the Novell Strategy
    Managing Software Distribution in a NetWare Environment
    NDS Planning Methodologies
    NetWare Distributed Management Services
    NetWare/IP: Overview, Planning, Installation, Configuration and
                Tuning
    NetWare 4 - Implementing Services and Clients
    Novell DOS 7 and Personal NetWare
    TUXEDO - The Key to Corporate Integration
    Understanding the NetWare Management System
    Understanding the Veritas File System
    UnixWare 1.1 Technical Overview
    UnixWare Device Drivers
    Visual AppBuilder, ALMs and AppWare Bus

Additional topics such as the NFS Product Family, NetWare Printing,
Tuning DOS and Windows Workstations as NetWare Clients and Global
Messaging are planned but as yet not confirmed.

Other sessions will be included from suppliers such as Lotus, concerning
Lotus Notes, and Saber Software, concerning the Saber Lan Workstation
product. It is hoped to have a parallel stream running through the
conference devoted to third party suppliers which complement the Novell
product lines.                

-----------------------------------------------------------------------

               Salford University Network Services
              Novell Networking Technical Conference

           University of Salford, 11th - 14th July 1994
                     Educational Booking Form

Please reserve a place at the Conference for:

Title:-

First Name:-

Surname:-

Address:-


Telephone:-

Fax Number:-

Email Address:-

Sex (M/F):-


Name and institution for the name badge:

Name:-

Position:-

Institution:-


Special requirements (diets etc):

Vegetarian:-

Vegan:-

Disabilities:-

Other:-


The Conference fee is œ450 which includes bed, breakfast, lunch, dinner, 
evening entertainment and all refreshments from lunch time on the 11th 
July to lunch on the 14th July. Free parking will be available at the 
University during the conference. Only eighty rooms are available with 
en-suite facilities (the remainder have communal facilities similar to most 
student residences). The en-suite rooms will be allocated on a first come 
first served basis so please book early.


Bed and Breakfast required for
the night of Sunday 10th July at          YES  /  NO
an additional cost of œ32.00

Simple evening meal required at 7.30pm     YES  /  NO
Sunday 10th July at an additional
cost of œ18


Address to which the invoice for the Conference fees should be sent:

Name:-

Address:-



Official Order Number:-


Transport to the railway station required on the 14th July   YES / NO

Transport to the airport required on the 14th July           YES / NO


Please note that no refund can be given on cancellations arriving less than
ten working days before the start of the Conference, or on failure to
attend.

For further details on the conference programme please contact Peter
Atkins on 061 745 5656.

The completed registration form should be returned to Peter Atkins,
either by Fax to 061 745 5888, by email to p.e.atkins@suns.salford.ac.uk 
or by post to, Peter Atkins, AIS Building, University of Salford, Salford, 
M5 4WT. The closing date for registrations is Friday 1st July. Please note 
that places will be limited, and will be filled on a first come, first 
served basis.

Full conference and arrival details will be sent to attendees during the
week before the conference.



-----------[000397][next][prev][last][first]----------------------------------------------------
Date:      23 May 1994 13:46:54 GMT
From:      bortz@cnam.cnam.fr (Stephane Bortzmeyer)
To:        comp.protocols.tcp-ip
Subject:   Re: MacTCP & SLIP Help!!

ftlofaro@unlv.edu (Frank Lofaro) writes:

>Bwahahahaha!!! PPP has _less_ overhead! Get real! SLIP just uses one
 ...
>SLIP is nice, efficient and simple. PPP adds all sort of link-level cruft, 
>and extra negotiation. With SLIP to just receive and send IP datagrams. With 

There are two issues involved here:
- wether PPP is better than SLIP or vice-versa
- wether the original poster who just wanted to connect to the Internet from
his/her Macintosh should use MacPPP or InterSLIP.

The reply to the first issue is not trivial and is open to debate (or
flame).
The reply to the second is far more simple: use MacPPP. It's fast, simple to
configure and incredibly robust. Of course, your service provider has to
support PPP (they should, IMHO).

>Oh, there is MUCH more chance of a buggy PPP implementation, it takes a 
>special breed of incompetant programmer to screw up SLIP.

May be in theory. But in practice (which was the concern of the original
poster), MacPPP (*one* specific implementation) is less buggy than any
*Macintosh* SLIP implementation.

>Check out the SLIP (1055) and PPP (I forget the number of the week, they 
>keep changing it) RFC's, and you'll see what I mean!

PPP is in RFC 1548.
You can also see discussions of the PPP vs. SLIP issue (these papers are as
one-minded as your post, albeit on the other side :-) on :
http://cs.uni-bonn.de:/ppp/faq.html
http://morningstar.com/

To quote the SLIP RFC (1055), which is an example of humility:

DEFICIENCIES

   There are several features that many users would like SLIP to provide
   which it doesn't.  In all fairness, SLIP is just a very simple
   protocol designed quite a long time ago when these problems were not
   really important issues.  The following are commonly perceived
   shortcomings in the existing SLIP protocol:

      - addressing:
...
      - type identification:
...
      - error detection/correction:
...
      - compression:
...

   Some work is being done by various groups to design and implement a
   successor to SLIP which will address some or all of these problems.

(End of quotation.)

From the point of view of the service provider (we have a small dialup IP
access at my university), I can say I appreciate the negociation features
of PPP (which are completely missing in SLIP). With SLIP, the only error
message one user get when misconfiguring its client is either "IP checksum
error" or total silence...

Stephane Bortzmeyer           Conservatoire National des Arts et Metiers	
bortzmeyer@cnam.fr            Laboratoire d'Informatique
                              292, rue Saint-Martin			
tel: +33 (1) 40 27 27 31      75141 Paris Cedex 03
fax: +33 (1) 40 27 27 72      France	

"C'est la nuit qu'il est beau de croire a la lumiere." E. Rostand

http://web.cnam.fr/personnes/bortzmeyer/home_page.dom

					
	




-----------[000398][next][prev][last][first]----------------------------------------------------
Date:      Mon, 23 May 94 21:20:00 -0400
From:      ivin.porteous@canrem.com (Ivin Porteous)
To:        comp.protocols.tcp-ip
Subject:   Setting up a small netwo

Hello All,

This might be a stupid question but I need to ask since I don't know the 
answer:

Here's the situation.... I would like to setup a small network using 
TCP/IP for OS/2 on the Server and TCP/IP for Dos/Windows on the 
Workgroup for Windows clients.  The questions is .. How do I go about 
assigning IP addresses?  I would also like to use TCP/IP to access the 
internet. How do I assign a unique ip address so that mail on the 
internet will be route correctly?


Any comments or suggestions would be very much appreciate,


Ivin

---
* CmpQwk #UNREG* UNREGISTERED EVALUATION COPY

-----------[000399][next][prev][last][first]----------------------------------------------------
Date:      Mon, 23 May 1994 15:54:23 GMT
From:      varney@cbnewsd.cb.att.com (Al Varney)
To:        comp.protocols.tcp-ip
Subject:   Re: CTRL-ALT-DEL problem with sockets betn. PC and Unix

In article <sivapras.769378914@shazam.cs.iastate.edu> sivapras@cs.iastate.edu (Gowri Sankar Sivaprasad) writes:
>Hello,
>
>	I am facing a problem with UNIX sockets.
>	
>	Problem:  I establish a TCP connection between my SCO UNIX system and a
>	Windows system (486 dx).  Then I reboot the windows m/c (CTRL_ALT_DEL).
>	But the SCO system does not recognize that the client has gone down. 

           That's because the SCO system had nothing to recognize.
           You need a timeout (aka "keep-alive") mechanism from the SCO end.

>	If I do a netstat on the SCO system, I find the socket to that client as
>	"ESTABLISHED".  This is a major problem because it eats up TCP ports.
>	
>	My Windows application is in Visual Basic.
                                     ^^^^^^^^^^^^^

   Is this the reason for all the reboots???

   This is also a problem when suddenly rebooting the SCO system,
   but I assume that is not necessary (except infrequently)?

Al Varney

-----------[000400][next][prev][last][first]----------------------------------------------------
Date:      Mon, 23 May 1994 16:38:50 GMT
From:      alfa@cix.compulink.co.uk ("Alfa Systems")
To:        comp.protocols.tcp-ip
Subject:   Re:   !!!

I got a whole load of non-ASCII characters plus the ones you got. I guess 
it makes more sense with a Cyrillic font :-)

Kin Ming Looi
Alfa Systems Limited
alfa@cix.compulink.co.uk

-----------[000401][next][prev][last][first]----------------------------------------------------
Date:      23 May 1994 16:42:57 GMT
From:      landmark@cs.tu-berlin.de (Torsten Kerschat)
To:        comp.unix.programmer,comp.unix.solaris,comp.unix.internals,comp.protocols.tcp-ip
Subject:   How to communicate with IP/UDP from Device Driver ?

Hello, (this question belongs to Solaris 5.3)
who can help me ? When I open a UDP socket with a socket call
and bind this to a port, I can link a device driver above
this socket. That's not a problem, I have done that and
I can communicate with the UDP modul.
What I want to do now i, to send a IP_ADD_MEMBERSHIP
request to the UDP (and underlying IP) modul from the 
linked device driver.
I can't do that with a setsockopt from above the device
driver. Who can tell me which kind of primitives
I have to use (I still have the Transport Provider
Interface Description) to do this  ?
Thanks in advance.

Torsten
-- 
Torsten Kerschat - Interdepartmental Research Center for Process Control
		   Technical University of Berlin (PRZ - Room HE 104)
Internet: torsten@prz.tu-berlin.d400.de / landmark@cs.tu-berlin.d400.de
Phone   : ++49 / (0)30 / 314 - 26822    Fax: ++49 / (0)30 / 314 - 21114

-----------[000402][next][prev][last][first]----------------------------------------------------
Date:      Mon, 23 May 1994 17:00:36 GMT
From:      rayman@csn.org (Ray Caruso)
To:        comp.protocols.tcp-ip
Subject:   Recieving ICMP Packets???

I have heard that it is possible to open a raw icmp socket and recieve 
all icmp traffic comming into the ICMP layer. 
So far, I am able to recieve any ICMP reply comming back into the system, 
but I do not get any ICMP requests.  
I am trying to build a proxy that will reply to ICMP echo Requests
targeted  for fake IP Address (read: no system has the address).
I have poked the ARP cache with the Fake IP Address and 
link level address of the system running my proxy. This way, IP packets
targeted for the fake IP Address are sent to the link level address of 
my Proxy host.

My code opens a raw icmp socket, and then select()'s on the file descriptor.
I have tried bind()'ing the socket to icmp and the fake IP address, but
that failed. 
Is there anything else I must do to get this to work?

Help. . .

I am running on SUN OS 4.1.3 and HP-UX 9.X.

Rayman
--
Ray Caruso
Power Play Technologies, Inc.
2231 Cedarwood Drive
Fort Collins, CO 80526
USA
303/227-1806
303/493-5168 FAX

-----------[000403][next][prev][last][first]----------------------------------------------------
Date:      Mon, 23 May 1994 17:10:31 GMT
From:      rayman@csn.org (Ray Caruso)
To:        comp.protocols.tcp-ip
Subject:   Recieving ICMP Packets???


Hello,

I have heard/read/dreamed that it is possible to open a raw
socket to icmp and recieve all incomming ICMP traffic.

I am trying to build an ICMP reply proxy for fake IP Address 
(read: no system on the net has the address, but we would like
to ping it).

I have written code that open a raw icmp socket and select()'s 
on it. The only data I seem to get are imcomming replies to 
ICMP requests from the local node. I need to recieve the 
ICMP requests that come into the local system.

I have poked the ARP cache so that the Fake IP Address in question
is matched up with the link level address of the node running my
proxy. This way, any IP packets targeted for the Fake IP Address
are sent the the link level address of my proxy. 

I want IP or ICMP to forward incomming ICMP traffic to my proxy.
What must I do . . . help. . . 

I am running on both SUN OS 4.1.3 and HP-UX 9.x

Thanx in advance

Rayman



--
Ray Caruso
Power Play Technologies, Inc.
2231 Cedarwood Drive
Fort Collins, CO 80526
USA
303/227-1806
303/493-5168 FAX

-----------[000404][next][prev][last][first]----------------------------------------------------
Date:      Mon, 23 May 1994 17:23:53 GMT
From:      jas@talking.COM (Jim Shankland)
To:        comp.protocols.tcp-ip
Subject:   Re: MacTCP & SLIP Help!!

In article <1994May23.050343.19139@unlv.edu> ftlofaro@unlv.edu (Frank Lofaro) writes:
[Anti-PPP diatribe:  SLIP is simpler, more reliable, doesn't add "cruft",
etc.]

PPP indeed has very slightly more overhead than SLIP -- mostly the CRC,
which SLIP omits.  And it is unquestionably more complicated to
*implement* -- not to use.  "Link-level cruft" is, of course, a
loaded term:  if you really need it, it's not cruft.

Now, here's the important part:  if you care about your data,
especially if you're going to be running Van Jacobson header
compression, you *really need* that link-level CRC.  V-J decompression
will happly turn utter gibberish into a perfectly reasonable-looking
TCP packet.  Most of the time, the TCP checksum will catch the
error; about 1 time out of 65536, it won't.

Needs vary.  The PPP implementations I've used have been free and
extremely robust.  The extra couple of bytes of overhead has
been unnoticeable.  And I've been thankful for the additional
reliability.

jas

-----------[000405][next][prev][last][first]----------------------------------------------------
Date:      23 May 1994 18:00:30 GMT
From:      eb@iunet.it (Enrico Badella)
To:        comp.protocols.tcp-ip
Subject:   Setting hop counts in ICMP packets


I'm reading the Stevens book and as an exercise I'm writing my own ping. What
I would like to do is to set the hop counts from a command line parameter, but
cannot find the correct option for setsockopt. I'm using SunOS 4.1.1 and Linux
1.1.0; looking around in the Linux kernel I saw that this option is settable
using IP_TTL but apparently nothing changes

Thank in advance.


-----------[000406][next][prev][last][first]----------------------------------------------------
Date:      23 May 1994 18:04:16 GMT
From:      jjlis@ict.pwr.wroc.pl (Jaroslaw Lis)
To:        comp.protocols.tcp-ip
Subject:   Re: Any Public Domain Simple router software available ?

Damiano Bolla (db15@ukc.ac.uk) wrote:
: I have to manage a simple network like this
 
: 	11	12	13
: 	|	|	|
: 	R1	R2	R3
: 	------------------- net 2 ------- cisco --- World
 
: 11,12,13,2 are subnets of a Class B network with netmask 255.255.255
: The routers (R) are Unix machines and have a static routing table
: to avoid problems at boot. (Some mounts directory from others)
: The cisco had a static routing table but now the admin wants
: to go dinamic I.e. routers R1,R2,R3 should tell to the cisco 
: that they can route to the given subnet. 
 
: The question is:
: Can I keep a static routing table into router R1,R2,R3 and 
: in the same time tell Cisco that they are routers for
: the given subnets ? 

Yes, it should work. One possible problem is static "default" routing
at R1,R2,R3. They can broadcast to all other host, that your Cisco
is 'default'.

Jarek.



-----------[000407][next][prev][last][first]----------------------------------------------------
Date:      23 May 94 19:35:14 GMT
From:      bas@moria.macs.sp.unisys.com (Brian Strop)
To:        comp.protocols.ppp,comp.protocols.tcp-ip
Subject:   NCSA 2.6 and InterSLIP problems


I have a copy of NCSA Telnet 2.6 and InterSlip 1.0.1.  I am connecting
to a Unisys ATS II terminal server (which is really a Xylogics Annex
terminal server under the hood).  I am running at 9.6Kbps and have
the RFC1044 compression option turned on in my InterSlip configuration.
I am running an MTU of 1006 bytes.  I am NOT using hardware handshaking (
I don't believe I have a cable to even do this).  I get connected to the
terminal server and can open a Telnet session, but most of the time, my
session will hang.  Sometimes I can do quite a few things (such as vi,
ls, pwd, etc. -- by the way, I am connecting to both Suns and Unisys
UNIX boxes), but eventually the session hangs.  If I establish another
session and perform a netstat on the UNIX system, the connection is
in the ESTABLISHED state, but there is data on the send queue.  Since
the session never times out, I assume the Mac is not in retransmission
mode (I am using MacTCP 2.0.4).  It almost appears as if the TCP window
is shutdown and not being re-opened.  Do you have any ideas?  If you
need any more information, I will gladly give it to you.  FTP does not
work either.  I can establish an ftp session, but most of the time, no
data (or maybe the very first block of data) gets transferred.  I am
trying to transfer an ASCII file.

I'd also like to find out why the session does not timeout?  Either TCP is
not retransmitting, or the data is being held for flow control reasons.  I
have MacTCP Watcher running.  It is not showing a re-transmission until I
close the session.  Does this mean the re-transmit was queued waiting for
some flow control condition to be cleared?

Any help would be greatly appreciated.

Thanks,

Brian
-- 
Brian A. Strop				(612) 687-2709
UNISYS					bas%moria.uucp@email.sp.paramax.com
Network Systems

-----------[000408][next][prev][last][first]----------------------------------------------------
Date:      Mon, 23 May 1994 19:55:02 GMT
From:      rstevens@noao.edu (W. Richard Stevens)
To:        comp.protocols.tcp-ip
Subject:   Re: HELP NEEDED: Setting timeout on sockets?

> The server waits for a connection from a client. As soon as a connection is 
> established,it immediately closes the connection and quits.

This causes the server's TCP to send a FIN to the client.  The client TCP
receives the FIN and says effectively "I won't receive any more data from
the other end, but I can still send data to the other end."  The reason
the client's TCP thinks it can still send data is because *all* a FIN
means is that the sender of the FIN is done sending.  Because of TCP's
half-close, the client's TCP lets the client process send data to the
server.

> The client establishes a connection with the server. It waits till the server
> quits ( I made it wait, of course!) and for three times it attempts to read from
> the server and three times to write something to server. These are my
> observations:
> 
> 	* Reads written with a return value 0. And the errno is set to "Error 0".
> 	  No signal is generated.

Exactly.  When you read from a socket that has received a FIN you get
an end-of-file.  Unix EOF is a read() returning 0.  I never said you
get the signal when reading, only when writing.

Also, please note that you *cannot* look at errno unless the function
returns an error.  That's why you always get "Error 0".

> 	* First write is successfull with a return value equal to the number of
> 	  bytes. errno is "Error 0". No signal generated.

Exactly.  Your TCP thinks it can still send data to the other end.  It
sends the data.  write() returns OK because all that happens is the
data is copied into your socket's TCP send buffer.  A successful return
from write() means very little in the networking world, because most
interesting errors happen at a later time (i.e., asynchronously).

When the server's TCP receives the data from the write(), it realizes that
the server process is gone, and responds with an RST.
 
> 	* Second write is unsuccessfull with a return value -1. errno is "broken
> 	   pipe". SIGPIPE is generated.
> 	* Same with the third write.

Exactly.  By the time you do the 2nd write(), the RST has been received.

> Now my questions are:
> 1. When is this RST sent?
> 2. Why didn't the read return an error with errno set to ECONNRESET?
> 3. Why was the first write successful?

See above.  The question "how do a get the first write after the other
end has terminated to generate SIGPIPE" shows up here constantly.  The
answer is "you can't."  If you want to know as soon as the process at
the other end of a connection terminates, use select(), testing for
readability, then the read will return 0 (as above).

	Rich Stevens

-----------[000409][next][prev][last][first]----------------------------------------------------
Date:      Mon, 23 May 94 21:08:49 GMT
From:      ftlofaro@unlv.edu (Frank Lofaro)
To:        comp.protocols.tcp-ip
Subject:   Re: HELP NEEDED: Setting timeout on sockets?

Here is an idea to avoid problems with not getting EPIPE/SIGPIPE until the 
second write. How about an option to have write() block until the data is 
either ACK'd by the remote side or a RST or other connection failure occurs. 

This could have negative performance implications however...
(there might be a way to fix those though)


-----------[000410][next][prev][last][first]----------------------------------------------------
Date:      Mon, 23 May 1994 22:13:28 GMT
From:      rstevens@noao.edu (W. Richard Stevens)
To:        comp.protocols.tcp-ip
Subject:   Re: Setting hop counts in ICMP packets

> I'm reading the Stevens book and as an exercise I'm writing my own ping. What
> I would like to do is to set the hop counts from a command line parameter, but
> cannot find the correct option for setsockopt. I'm using SunOS 4.1.1 and Linux
> 1.1.0; looking around in the Linux kernel I saw that this option is settable
> using IP_TTL but apparently nothing changes

The TTL field is hard to set from a process.  Plus it depends on your
operating system, as this has changed over time.  I've never seen Linux,
so I have no idea what it does.

All this aside, I see in 4.4BSD that the IP_TTL socket option only affects
TCP and UDP sockets.  For raw sockets without the IP_HDRINCL socket option
you always get 255.  If you want to set your own TTL for a raw socket you
need to write your own IP headers (IP_HDRINCL) as traceroute does.

	Rich Stevens

-----------[000411][next][prev][last][first]----------------------------------------------------
Date:      Tue, 24 May 94 07:50:00 PDT
From:      adar0@routers.com
To:        comp.protocols.tcp-ip,comp.protocols.ppp
Subject:   Re: NCSA 2.6 and InterSLIP problems



> I have a copy of NCSA Telnet 2.6 and InterSlip 1.0.1.  I am connecting
> to a Unisys ATS II terminal server (which is really a Xylogics Annex
> terminal server under the hood).  I am running at 9.6Kbps and have
> the RFC1044 compression option turned on in my InterSlip configuration.
> I am running an MTU of 1006 bytes.  I am NOT using hardware handshaking 

Brian,
I've seen this happen on other machines before and every time it has
been related to a hardware handshaking problem; or, if you really
expect to use software flow control, one or more components in the
end-to-end loop is not playing by the same rules (eg, xon/xoff is
intercepted by a modem for example and not passed on to the piece
of equipment that you expected to deal with it).
Your session is likely in a state where flow control has stopped
data in the middle of a packet.  Packet handlers typically do not
time out and request a retransmission when they have been "flow
controlled".
Rich
adar0@routers.com


-----------[000412][next][prev][last][first]----------------------------------------------------
Date:      Tue, 24 May 94 08:05:23 PDT
From:      adar0@routers.com
To:        comp.protocols.tcp-ip
Subject:   Re: Assigning IP addresses to a CISCO WAN


> We currently
> are having a discussion about how to assign IP numbers to the WAN
> Synch ports.
> 
> Each LAN has its own class C IP subnet, and currently the two 
 synchronous
> ports comprise a third class C subnet. 
> 
Mark,
Technically, either solution will work and will not cause any long
term problems.  However, business growth and/or practicality may
impact which approach is more appropriate.  From a business perspective,
if the business grows, using a Class-C for the backbone net has the
advantage of being able to be "managed" by a separate group while 
the end Class-C's are managed by their respective groups.  The
disadvantage is that you cannot break ONE class-c address space into
two geographic areas and expect the sessions to function through a
second class-c address space.  All addresses within a class must be
contiguous.
From a practicality perspective, assuming the business does not grow
much, burning a full class-c for a couple of serial links is quite
wastefull.
Rich
adar0@routers.com


-----------[000413][next][prev][last][first]----------------------------------------------------
Date:      Tue, 24 May 94 08:15:11 PDT
From:      adar0@routers.com
To:        comp.windows.x,comp.unix.misc,comp.sys.novell,comp.protocols.tcp-ip
Subject:   Re: Connecting PCs & wkstations...



> We've been trying to connect our two networks, token-ring and
> ethernet network. 
 (stuff deleted)
> Did anybody experienced such a problem? Where seems to be a problem?
> On workstations, with Chameleon or with the LAN server?
> 

I cannot tell for sure whether this is the same problem that I've
seen before, but quite likely it is.  It is likely related to the
differences in how token ring and ethernet handle the mac layer or
hardware addresses.  There are slight differences between the two and
you probably won't see it unless you use a sniffer (or some other
diagnostic tracing tool) to look at the individual packets.  The
problem relates to two items:  1) when IBM popularized token ring,
their design shifted the hardware address onto the wire in reverse
bit order from ethernet standards, and, 2) some software implementations
use a couple of bits in the hardware address to mean special things
like group addressing, user administered addressing and multicasting.
If the X software that you're using makes use of these two bits, it
is highly likely the software that is performing the token ring to
ethernet bridging is not handling these two bits properly.  Or, saying
it another way, the software is not translating the two bits from
token ring into the appropriate two bits on ethernet.
Rich
adar0@routers.com


-----------[000414][next][prev][last][first]----------------------------------------------------
Date:      24 May 1994 01:31:46 GMT
From:      ccetc@nusunix1.nus.sg (SuperFast)
To:        comp.protocols.tcp-ip
Subject:   Two NIC in the same ethernet segment, possible?


Is it possible to put two Ethernet cards in a UNIX box, and assign them
two different IP address for the same subnet?

The intension is to split network traffics/load between the two cards so that
higher performance can be achived. The other possibility is to put a UNIX server
machine with two NICs to a EtherSwitch so that the total bandwidth
become 2x10Mbps.

Is this possible at all?

--
ccetc@leonis.nus.sg
National University of S'pore

-----------[000415][next][prev][last][first]----------------------------------------------------
Date:      Tue, 24 May 94 08:44:21 PDT
From:      adar0@routers.com
To:        comp.protocols.tcp-ip
Subject:   Re: IGRP proprietary



> Hi, I need to know if the IGRP protocol (CISCO's routing protocol used 
> in internet) is proprietary of CISCO or other manufacturers handle that.
> Please send answers (if any) to villa@dsi.unimi.it.
> Vilas

It is Cisco's proprietary routing protocol.  I do not know of any
other vendor supporting this protocol, although you might get a
better answer by reposting this question in /comp/dcom/sys/cisco.
They generally have several technical and management people watching
and responding to articles in that group.


-----------[000416][next][prev][last][first]----------------------------------------------------
Date:      Tue, 24 May 94 08:48:39 PDT
From:      adar0@routers.com
To:        comp.protocols.tcp-ip
Subject:   Re: KA9Q with dial-in slip.



> I wan't the SLIP connections to have a completly different address.
> This way I don't have to register it with our network people.  Is it
> possible to do this with out changing anything in our existing
> network.  Is it possible to telnet to IP address 141.202.200.1 from a
> slip connection IP address 141.200.200.1.  Does anyone have any ideas?

It may be possible to accomplish what you're trying to do, however the
value of it seems useless.  Assuming you got everything to work and
your system admin folks don't have anything installed to stop you,
when you telnet from the fictitious domain into the company domain your
fictitious address will show up in company hosts, logs, etc.  So it
will just be a matter of time before they see what your doing.  You
can judge what the consequences will be.  Based on how you asked the
question, I'd suspect it will take quite abit of your time to get it
to work.

-----------[000417][next][prev][last][first]----------------------------------------------------
Date:      24 May 1994 02:01:41 GMT
From:      nhs@llnl.gov (Norman H. Samuelson)
To:        comp.protocols.tcp-ip.ibmpc,comp.protocols.tcp-ip,comp.protocols.ppp
Subject:   Re: security of SLIP/PPP

In article <ZHAO.94May23150456@sparta.crl.nmsu.edu> zhao@crl.nmsu.edu (Z. Zhao) writes:
>I want to connect some PCs in my company to a SPARC mail-server by
>using SLIP/PPP. My boss asks me: "Is it secure?" I cannot answer. 
>
>I assume there was no firewall installed on the mail-server. Is
>SLIP/PPP connection are more secure, or there are easier solution on
>security issues, comparing with ethernet connection?
>
>
>ZiZi

Is your current system secure?  Do you allow dialup access?  Are you
connected to the internet?  If you are already connected to the outside
world, either by dialup or the internet, adding SLIP or PPP should not
make much difference.  It is somewhat more likely to affect the security
of the systems which dial in to the net, unless the users are careful with
such things as ftp servers.

- Norm -

-----------[000418][next][prev][last][first]----------------------------------------------------
Date:      Tue, 24 May 94 09:14:00 PDT
From:      adar0@routers.com
To:        comp.protocols.tcp-ip
Subject:   Re: Two NIC in the same ethernet segment, possible?



> Is it possible to put two Ethernet cards in a UNIX box, and assign them
> two different IP address for the same subnet?
> 
> The intension is to split network traffics/load between the two cards so 
 that
> higher performance can be achived. The other possibility is to put a 
 UNIX server
> machine with two NICs to a EtherSwitch so that the total bandwidth
> become 2x10Mbps.
> 
> Is this possible at all?
> 

Sure is possible, either on the same subnet or on different ones. I
can not comment on any improvements on performance though.  I'd suspect
that if you have a performance bottleneck with the single ethernet
adapter, installing a second one on the same subnet or wire isn't
likely to improve much due to other limitations including internal
unix code, routers/bridges, and the like.


-----------[000419][next][prev][last][first]----------------------------------------------------
Date:      Tue, 24 May 1994 02:34:42 GMT
From:      gwright@world.std.com (Gary R Wright)
To:        comp.protocols.tcp-ip
Subject:   Re: Recieving ICMP Packets???

In article <Cq9Lx0.9ux@csn.org>, Ray Caruso <rayman@csn.org> wrote:
>I have heard that it is possible to open a raw icmp socket and recieve 
>all icmp traffic comming into the ICMP layer. 
 
>So far, I am able to recieve any ICMP reply comming back into the system, 
>but I do not get any ICMP requests.  
 [...]
>Is there anything else I must do to get this to work?
 [...]
>I am running on SUN OS 4.1.3 and HP-UX 9.X.

On any system derived from the BSD networking code, you will not be able
to receive ICMP requests on a raw ICMP socket.  The kernel handles
requests directly calling icmp_reflect() to send the correct ICMP reply.
For other ICMP packets (replies and errors) the packet will be made 
available to raw ICMP listeners as you described.

There are two solutions:
	- modify the kernel to pass requests to the raw socket
	- use a packet sniffer (e.g., Berkely Packet Filter, STREAMS packet filter) 
	  to capture the incoming packet (you would still have problems with the 
	  kernel responding to the packet)

-----------[000420][next][prev][last][first]----------------------------------------------------
Date:      24 May 94 03:55:42 GMT
From:      rtkao@remus.rutgers.edu (Richard Kao)
To:        comp.protocols.tcp-ip,comp.protocols.tcp-ip.ibmpc
Subject:   Re: KA9Q with dial-in slip.

Both of these domains are not connected to the internet.  They are my
companies own isolated networks.  There is no connection to the
internet for security purposes.

-----------[000421][next][prev][last][first]----------------------------------------------------
Date:      24 May 1994 16:31:11 -0700
From:      mike@software.com (Michael D'Errico)
To:        comp.protocols.tcp-ip
Subject:   Re: Pushing data back onto my socket connection?

jeff@astph (Jeff Martin) writes:

>I have some memory from my reading of being able to read 
>from a socket and optionally push some of this data back
>onto the socket connection, perhaps to be read in again
>by another server process that has access to the same
>connection.

Not exactly.  It is possible to peek at the pending data
without affecting the buffer, but not to push it back.
Instead of using read() you need to use recv() with the
MSG_PEEK flag set.

Michael D'Errico
mike@software.com


-----------[000422][next][prev][last][first]----------------------------------------------------
Date:      Tue, 24 May 1994 07:11:43 GMT
From:      logtel@zeus.datasrv.co.il (LOGTEL)
To:        comp.protocols.tcp-ip
Subject:   Fine-tuning TCP/IP

Hi, TCP/IP gurus

Is it really possible to somehow fine-tune the packet size for the
UNIX TCP/IP implementations ?

Simha Karlin, LogTel Systems
email: logtel@datasrv.co.il

-----------[000423][next][prev][last][first]----------------------------------------------------
Date:      Tue, 24 May 1994 08:07:39 GMT
From:      cpm@uniplex.co.uk (Chris P Murray)
To:        comp.protocols.tcp-ip
Subject:   Re: Help with BSD sockets

Ran-Chi Huang (rc@netarch.com) wrote:
: I am writing a client server application and it essentially works except
: for one minor problem.
 
: When I kill the daemon appsd via "kill -9 17897", the daemon indeed
: goes away.
 
:     root 17897   111 80 23:49:02 ?          0:03 appsd
 
: However, when I try starting it up again, I cannot start it. Upon
: checking I find the following lying around when I do a "netstat -a".
: After I wait for 3-4 minutes, these TIME_WAIT connections disappear
: and I am able to start the daemon again.

(stuff deleted)

Set the SO_REUSEADDR socket option on your server's listening socket.


-----------[000424][next][prev][last][first]----------------------------------------------------
Date:      Tue, 24 May 1994 15:39:32
From:      timc@sni.com.au (Tim Cullen)
To:        comp.protocols.tcp-ip
Subject:   Non-Interactive input to telnet

This question has probably been already addressed in a FAQ.

How can I use telnet with input from a file (on a Unix host) ?

i.e telnet host < cmdlist
Regards,

Tim Cullen (timc@sni.com.au)
Unix Product Manager
Siemens Nixdorf Information Systems Pty Ltd
655 Pacific Highway, St Leonards, NSW, 2065, AUSTRALIA
Voice: +61-2-430-2154, Fax: +61-2-439-5734

-----------[000425][next][prev][last][first]----------------------------------------------------
Date:      24 May 1994 12:58:52 GMT
From:      jim@cs.strath.ac.uk (Jim Reid)
To:        comp.protocols.tcp-ip
Subject:   Re: Any Public Domain Simple router software available ?

In article <2rqr70$qoi@sun1000.ci.pwr.wroc.pl> jjlis@ict.pwr.wroc.pl (Jaroslaw Lis) writes:

   Damiano Bolla (db15@ukc.ac.uk) wrote:
   : I have to manage a simple network like this
 
   : 	11	12	13
   : 	|	|	|
   : 	R1	R2	R3
   : 	------------------- net 2 ------- cisco --- World
 
   : 11,12,13,2 are subnets of a Class B network with netmask 255.255.255
   : The routers (R) are Unix machines and have a static routing table
   : to avoid problems at boot. (Some mounts directory from others)
   : The cisco had a static routing table but now the admin wants
   : to go dinamic I.e. routers R1,R2,R3 should tell to the cisco 
   : that they can route to the given subnet. 
 
   : The question is:
   : Can I keep a static routing table into router R1,R2,R3 and 
   : in the same time tell Cisco that they are routers for
   : the given subnets ? 

   Yes, it should work. One possible problem is static "default" routing
   at R1,R2,R3. They can broadcast to all other host, that your Cisco
   is 'default'.

That is the wrong answer unfortunately. The routers will have to
advertise routes to their networks to each other and to the Cisco.
There are two options. One is for the Cisco and the routers to run a
routing daemon to exchange information with each other. The other is
for the Cisco and routers to configure static routes for themselves.

-----------[000426][next][prev][last][first]----------------------------------------------------
Date:      Tue, 24 May 1994 13:10:25 GMT
From:      118980@canberra.rz.fh-offenburg.de (Roland Baer)
To:        comp.protocols.tcp-ip
Subject:   HELP messagequeus with pc/tcp

I have to port the messageques from unix(msg.h) with the pc/tcp developper 
kit to the dos world. 
The unix c-programs have to be source-code-compatible.
HELP ME!
roland f. baer
prog10@fh-offenburg.de

-----------[000427][next][prev][last][first]----------------------------------------------------
Date:      Tue, 24 May 1994 13:19:51 GMT
From:      clintp@world.std.com (Clinton A Pierce)
To:        comp.protocols.tcp-ip,comp.unix.questions
Subject:   Sockets connect(), what permissions involved?

Sorry about the crosspost--but this has been on the Net for a week and no-one
has taken a stab at it.

I'm tinkering with sockets on serveral AIX RS/600's and I have a short demo
program (enclosed).  I have one machine that will NOT run this code.  When
the client tries to connect(), he gets "Permission denied".   What permission?
When connect()ing to a socket, what permissions are involved?!?  I Can't find
any references to permissions in any FM.

Code edited for brevity:
The server's code

#include <sys/types.h>
#include <sys/socket.h>
#include <netinet/in.h>
#include <unistd.h>
#include <string.h>
#include <netdb.h>
#include <stdio.h>

extern int errno;

main(argc, argv)
int argc;
char **argv;
{
	char c;
	int sock;
	FILE *fp, *tempfile;
	int fromlen;
	char hostname[64], xms[123];
	struct hostent *hp;
	int i, s, ns;
	struct sockaddr_in sin, fsin;

	if (argc!=3) {
		fprintf(stderr, "Usage server socket filename");
	}
	tempfile=fopen(argv[2], "r");
	if (tempfile==NULL) exit(2);
	fclose (tempfile);
	sock=atoi(argv[1]);
	if (sock==0) exit(2);

	gethostname(hostname, sizeof(hostname));
	
	if ((hp=gethostbyname(hostname))==NULL) {
		fprintf(stderr, "%s: host unknown.\n", hostname);
		exit(1);
	}

	if ((s=socket(AF_INET, SOCK_STREAM, 0))<0) {
		perror("server:socket");
		exit(1);
	}

	sin.sin_family=AF_INET;
	sin.sin_port=htons(sock);
	bcopy(hp->h_addr, &sin.sin_addr, hp->h_length);

	if (bind(s, &sin, sizeof(sin)) < 0) {
		perror("server:bind");
		exit(1);
	}

	if (listen(s, 5)<0) {
		perror("server:listen");
		exit(1);
	}

	if ((ns=accept(s, &fsin, &fromlen))<0) {
		perror("server:accept");
		exit(1);
	}
	
	fp=fdopen(ns, "r");

[Useful portion of code removed]

	fclose(tempfile);
	close(s);
	exit(0);
}
	

And the client:

#include <sys/types.h>
#include <sys/socket.h>
#include <netinet/in.h>
#include <netdb.h>
#include <stdio.h>

#define NSTRS	3

char *strs[NSTRS] = {
	"This is first from the client\n",
	"This is second from the client\n",
	"This is third from the client\n"
};

extern int errno;

main(argc, argv)
int argc;
char **argv;
{
	char c;
	int sock;
	FILE *fp;
	int fromlen;
	char hostname[64];
	struct hostent *hp;
	int i, s;
	struct sockaddr_in sin;

	sock=atoi(argv[1]);
	if (sock==0) exit(2);

	gethostname(hostname, sizeof(hostname));
	
	if ((hp=gethostbyname(hostname))==NULL) {
		fprintf(stderr, "%s: host unknown.\n", hostname);
		exit(1);
	}

	if ((s=socket(AF_INET, SOCK_STREAM, 0))<0) {
		perror("client:socket");
		exit(1);
	}

	sin.sin_family=AF_INET;
	sin.sin_port=htons(sock);
	bcopy(hp->h_addr, &sin.sin_addr, hp->h_length);

	if (connect(s, &sin, sizeof(sin)) < 0) {
		perror("client:connect");
		exit(1);
	}

	fp=fdopen(s, "r");

	for(i=0; i<NSTRS; i++) {
		while ((c=fgetc(fp))!=EOF) {
			putchar(c);
			if (c=='\n') break;
		}
	}

	for(i=0; i<NSTRS; i++) 
		send(s, strs[i], strlen(strs[i]), 0);
	
	close(s);
	exit(0);
}
	


-----------[000428][next][prev][last][first]----------------------------------------------------
Date:      Tue, 24 May 1994 13:25:22 GMT
From:      erdmana@lbm.com (Alan Erdman)
To:        comp.protocols.tcp-ip
Subject:   queing with listen()

Hi,

   This is my first post - hope I'm giving enough info.

   I discovered this problem while using a wrapper for sockets called Simple Socket
Lib*ary from NASA, which uses connection-oriented stream sockets. The same thing 
occured using 'straight' BSD connection-oriented stream sockets.  Both were run 
under SunOS 4.1.3.
   
   The problem is that when requests that were qued on the socket (after setting with listen()) are accepted, they're received in reverse order (eg. incomming messages
1,2,3,4 will be read in the order 4,3,2,1).  Do I need to build a stack handler or
is there some way to remedy this with the existing socket software?  


Thanx in advance,
  
   Alan


----------------------------------------------------
Alan Erdman 		erdmana@lbm.com			    WANTED:
LB&M Associates  	(405) 581-3776 		          clever quote		
----------------------------------------------------


-----------[000429][next][prev][last][first]----------------------------------------------------
Date:      Tue, 24 May 1994 13:33:19 GMT
From:      ihsan@nmti.com (jaleel ihsan)
To:        comp.protocols.tcp-ip
Subject:   Re: Socket read question

In article <2rmnedINNglk@early-bird.think.com> barmar@think.com (Barry Margolin) writes:
>In article <id.W3O91.ENB@nmti.com> ihsan@nmti.com (jaleel ihsan) writes:
>>On a Posix compliant system, if I make a blocking socket read request for N
>>bytes, is the system required to block until at least N bytes are received,
>
>Read() on a socket is the same as read() on any other device.

Which means read N bytes until and unless some thing abnormal occurs, eg,
connection closed (EOF), OOB/PUSH flag received ?

What other situations will return < N bytes on BSD type socket ?  This was
the original intent of my question and I have not heard any answers yet !

>The size argument is just a maximum, not a requirement.

It is a requirement for maximum ie a read NEVER gives back > N bytes.

>Stream sockets are byte -- they don't include message boundaries.

Sorry if I implied otherwise.

>Consider that if you
>didn't know the size of the incoming message you would presumably provide a
>huge buffer, which would be pretty troublesome if it blocked until that
>amount were read.
>

Precisely the reason for my question.  If I requets N bytes, I expect N bytes
unless something unusual happens.  I am trying find out the complete definition
of "unusual" for sockets on a posix system.

Thanks for you help.

-----------[000430][next][prev][last][first]----------------------------------------------------
Date:      Tue, 24 May 1994 13:53:22 GMT
From:      mark@alias.com (Mark Andrews)
To:        comp.protocols.tcp-ip
Subject:   Re: How does a server detect a client crash using sockets?

In <1994May20.142214.1412@noao.edu> rstevens@noao.edu (W. Richard Stevens) writes:

>> Is there any way, within TCP, to know that the
>> process at the other end of a socket has crashed and left unread stuff
>> in a buffer?
 
>Not that I'm aware of.  You can tell that the other end has crashed (see
>my previous posting), but you have no information as to how much data
>is discarded on both ends when this happens.  (I can imagine some groveling
>through some kernel data structures to find out some information; for example
>your send buffer sizes give you some information as to what's been sent;
>TCP's sequence numbers tell you what's been sent and ACK'ed; the other
>side's window advertisements tell you how much data the application has
>read; etc.) but that seems pretty extreme.

What about ioctl(s, FIONREAD, &var)? That should give you some numbers
from the side of the connection that has not crashed, but probably not
any data in the  TCP window.

-----------[000431][next][prev][last][first]----------------------------------------------------
Date:      24 May 1994 14:14:48 GMT
From:      casper@fwi.uva.nl (Casper H.S. Dik)
To:        comp.protocols.tcp-ip
Subject:   Re: Socket read question

ihsan@nmti.com (jaleel ihsan) writes:

>In article <2rmnedINNglk@early-bird.think.com> barmar@think.com (Barry Margolin) writes:
>>In article <id.W3O91.ENB@nmti.com> ihsan@nmti.com (jaleel ihsan) writes:
>>>On a Posix compliant system, if I make a blocking socket read request for N
>>>bytes, is the system required to block until at least N bytes are received,
>>
>>Read() on a socket is the same as read() on any other device.
 
>Which means read N bytes until and unless some thing abnormal occurs, eg,
>connection closed (EOF), OOB/PUSH flag received ?

No.  Block until some bytes available or an error occurs.

>What other situations will return < N bytes on BSD type socket ?  This was
>the original intent of my question and I have not heard any answers yet !

When there are bytes avaiable, but < N bytes.

>Precisely the reason for my question.  If I requets N bytes, I expect N bytes
>unless something unusual happens.  I am trying find out the complete definition
>of "unusual" for sockets on a posix system.

This is not true for devices and sockets, only for plain files.

Casper

-----------[000432][next][prev][last][first]----------------------------------------------------
Date:      Tue, 24 May 1994 14:30:26 GMT
From:      wilfred@zeus.wnc.nedlloyd.nl (Wilfred Mollenvanger)
To:        comp.protocols.tcp-ip.ibmpc,comp.protocols.tcp-ip,comp.protocols.ppp
Subject:   Re: security of SLIP/PPP

Norman H. Samuelson (nhs@llnl.gov) wrote:
: In article <ZHAO.94May23150456@sparta.crl.nmsu.edu> zhao@crl.nmsu.edu (Z. Zhao) writes:
: >I want to connect some PCs in my company to a SPARC mail-server by
: >using SLIP/PPP. My boss asks me: "Is it secure?" I cannot answer. 
: >
: >I assume there was no firewall installed on the mail-server. Is
: >SLIP/PPP connection are more secure, or there are easier solution on
: >security issues, comparing with ethernet connection?
: >
: >
: >ZiZi
 
: Is your current system secure?  Do you allow dialup access?  Are you
: connected to the internet?  If you are already connected to the outside
: world, either by dialup or the internet, adding SLIP or PPP should not
: make much difference.  It is somewhat more likely to affect the security
: of the systems which dial in to the net, unless the users are careful with
: such things as ftp servers.
 
: - Norm -

You could also use the tcp wrapper program on your sparc to filter
incoming connections on source address and/or protocol type.
(cert.sei.cmu.edu:/pub/network_tools/tcp_wrapper.shar I believe)

Wilfred Mollenvanger.

-----------[000433][next][prev][last][first]----------------------------------------------------
Date:      Tue, 24 May 1994 14:35:22 GMT
From:      jeff@astph (Jeff Martin)
To:        comp.protocols.tcp-ip
Subject:   Pushing data back onto my socket connection?

I have some memory from my reading of being able to read 
from a socket and optionally push some of this data back
onto the socket connection, perhaps to be read in again
by another server process that has access to the same
connection.

For example several processes are all polling on many open
socket connections.  One server reads in 32 bytes of data,
but decides he cannot process the data yet and so wishes
to push the data back onto the connection to allow another
server to read it in at a later time.

Is this possible with the TCP/IP protcol?  And if it is possible
is it of any greater expense than normal socket communication?

Thanks, Jeff
-- 
Jeff Martin, dbms programmer,		Philadelphia Phillies
INET:	astph!jeff@cse.psu.edu		Voice:	(814)234-8592x32
UUCP:	psuvax1!astph!jeff		FAX:	(814)234-1269
SLOW:	141 West Beaver, Suite A, State College, PA 16801

-----------[000434][next][prev][last][first]----------------------------------------------------
Date:      Tue, 24 May 1994 16:18:30 GMT
From:      edgar@wired.com (Edgar Nielsen)
To:        comp.protocols.tcp-ip,comp.protocols.ppp
Subject:   Re: NCSA 2.6 and InterSLIP problems

adar0@routers.com wrote:


: > I have a copy of NCSA Telnet 2.6 and InterSlip 1.0.1.  I am connecting
: > to a Unisys ATS II terminal server (which is really a Xylogics Annex
: > terminal server under the hood).  I am running at 9.6Kbps and have
: > the RFC1044 compression option turned on in my InterSlip configuration.
: > I am running an MTU of 1006 bytes.  I am NOT using hardware handshaking 
 
: Brian,
: I've seen this happen on other machines before and every time it has
: been related to a hardware handshaking problem; or, if you really
: expect to use software flow control, one or more components in the
: end-to-end loop is not playing by the same rules (eg, xon/xoff is
: intercepted by a modem for example and not passed on to the piece
: of equipment that you expected to deal with it).

I'm having similar problems connecting with a Livingston router.
But the problem only happens if I telnet to a sun running solaris 2.3. I can't
get it to happen with either of our 2 other (non-sun)unix hosts. I
don't think it is a flow control problem since I can be using mosaic
and do massive ftp stuff from any other site with no problems.
But the telnet to the sun will hang everytime on doing little
things like during 'ps' output, 'ls', etc!

If I connect via ARA, NCSA telnet 2.6 has no such problems. I'm going to try
using telnet 2.5 to see if the problem still exists....

Edgar Nielsen

-----------[000435][next][prev][last][first]----------------------------------------------------
Date:      Tue, 24 May 1994 16:21:10 GMT
From:      leff@sco.com (Bill Leff)
To:        comp.protocols.tcp-ip
Subject:   ntp time servers


Can someone please email me the address of an ntp time
server on the internet? 

Thanks

-leff@sco.com


-- 
-------------------------
"I consider myself and still do consider myself the hippest man on the planet...
But I think if you have to say that you probably aren't, so I've never really
said that." -Barry Manilow

-----------[000436][next][prev][last][first]----------------------------------------------------
Date:      24 May 1994 16:25:28 GMT
From:      barmar@think.com (Barry Margolin)
To:        comp.protocols.tcp-ip
Subject:   Re: How does a server detect a client crash using sockets?

In article <Cq2uzI.KJs@sgi1.fnet.cs.mci.com> abell@velveeta.apdev.cs.mci.com (Andrew_Bell) writes:
>This I understand.  My question is what happens to things in buffers
>when a process goes down?  Certainly the write() call will succeed even
>if a read() has not yet occurred at the other end.  Is there a way to know
>that the process at the other end of a socket connection has read the
>data that TCP has sent?  Is there any way, within TCP, to know that the
>process at the other end of a socket has crashed and left unread stuff
>in a buffer?  How can the sending process know what stuff never got read
>and needs to be retransmitted when a new connection is made?  Does this
>have to be done within the application (client and server)?  Does any
>UNIX OS with asynchronous I/O help the situation (write now and tell me
>when the read happened later)?

There's nothing in the TCP protocol that provides any information about
which bytes have been read by the remote process.  The remote TCP should
acknowledge data as soon as it is put into a buffer, at which time it is
available for the remote process to read.  But if you need to know that the
remote process has actually read it, you need something at the application
level.  An example of this is TELNET's "AYT" (Are You There) control
sequence, which requires the remote TELNET to send back a string.
-- 
Barry Margolin
System Manager, Thinking Machines Corp.

barmar@think.com          {uunet,harvard}!think!barmar

-----------[000437][next][prev][last][first]----------------------------------------------------
Date:      24 May 1994 16:29:28 GMT
From:      nelson@crynwr.crynwr.com (Russell Nelson)
To:        comp.protocols.tcp-ip
Subject:   Re: Two NIC in the same ethernet segment, possible?

In article <2rrle2$ltt@nuscc.nus.sg> ccetc@nusunix1.nus.sg (SuperFast) writes:

   Is it possible to put two Ethernet cards in a UNIX box, and assign them
   two different IP address for the same subnet?

That won't help.  One card is sufficient to fill a 10Mbps network.  If
you can't do that, then the bottleneck is not at the card, it's likely
somewhere else.

   The intension is to split network traffics/load between the two
   cards so that higher performance can be achived. The other
   possibility is to put a UNIX server machine with two NICs to a
   EtherSwitch so that the total bandwidth become 2x10Mbps.

Better idea: Buy one of those new server-integrated switches, and get
60Mbps.  You'll have to write a Unix device driver for it, but they'll
likely help you because both companies doing it (so far) are startups
and are probably cash-poor at this point.

Xnet Technology, +1-408-263-6888, +1-408-263-8898 FAX, contct is Phil Edholm 

Xedia Corporation, +1-508-658-7200, info@xedia.com, contact is James P. Skinner

--
-russ <nelson@crynwr.com>      ftp.msen.com:pub/vendor/crynwr/crynwr.wav
Crynwr Software   | Crynwr Software sells packet driver support | ask4 PGP key
11 Grant St.      | +1 315 268 1925 (9201 FAX)    | Quakers do it in the light
Potsdam, NY 13676 | LPF member - ask me about the harm software patents do.

-----------[000438][next][prev][last][first]----------------------------------------------------
Date:      Tue, 24 May 1994 16:35:48 GMT
From:      hasty@netcom.com (Amancio Hasty Jr)
To:        comp.windows.x,comp.unix.misc,comp.sys.novell,comp.protocols.tcp-ip
Subject:   Re: Connecting PCs & wkstations...

In article <2rsvbj$207@ace.mid.net> adar0@routers.com writes:
>
>
>> We've been trying to connect our two networks, token-ring and
>> ethernet network. 
 (stuff deleted)
>> Did anybody experienced such a problem? Where seems to be a problem?
>> On workstations, with Chameleon or with the LAN server?
>> 

If it helps, while I was working at Cisco using their routers, I didn't
any experience any problems using X. My workstation was a PC running
FreeBSD.

You can sort out the problems with the Berkeley Packet Filter (bpf), 
etherfind   or a sniffer.

Amancio


-- 
FREE unix, gcc, tcp/ip, X, open-look, netaudio,  tcl/tk, MIME, midi,sound
at  freebsd.cdrom.com:/pub/FreeBSD
Amancio Hasty,  Consultant |
Home: (415) 495-3046       |  
e-mail hasty@netcom.com	   |  ftp-site depository of all my work:    
                           |  sunvis.rtpnc.epa.gov:/pub/386bsd/X

-----------[000439][next][prev][last][first]----------------------------------------------------
Date:      24 May 1994 16:36:02 GMT
From:      vishwamber.yelsangikar.1@nd.edu (Vishwamber Yelsangikar)
To:        comp.protocols.tcp-ip
Subject:   TCP Connection.

If a physical link is down and there is already a TCP connection what will
happen to that connection?  In my case when our router went down all my
remotely logged in windows disappeared.  When does TCP connection gets
disconnected under such circumstances? and how? gracefully or not? 
Logically there should never be any kind of disconnection.  It should get
hung during that extended period of router downtime.



Vishwamber Yelsangikar
 

-----------[000440][next][prev][last][first]----------------------------------------------------
Date:      24 May 1994 16:37:58 GMT
From:      scoggin@delmarva.com (John K Scoggin Jr)
To:        comp.protocols.tcp-ip
Subject:   Re: ntp time servers

In article 22078@sco.com, leff@sco.com (Bill Leff) writes:
> 
> Can someone please email me the address of an ntp time
> server on the internet? 
> 
> Thanks
> 

Retrieve the file clock.txt in the pub/ntp/docs directory on louie.udel.edu.
It has the addresses of all of the public access NTP time servers.

While you are at it, grab a copy of xntp3 and send it to your software development
folks!  I would really like to have an up-to-date version of xntp on my Open
Desktop systems...  :-)

	- John

---
+---------------------------------------------------------------------+    
|  John K. Scoggin, Jr.			Email: scoggin@delmarva.com   |
|  Supervisor, Network Operations              scoggin@ee.udel.edu    |
|  Delmarva Power & Light Company       Phone: (302) 451-5200         |
|  500 N. Wakefield Drive               NOC:   (800) 388-7076         |
|  Newark, DE 19714-6066		Fax:   (302) 451-5321         |
|  					Ham:   N3??? (real soon now!) |
|  The opinions expressed are not those of Delmarva Power, simply the |
|  product of an over-active imagination...                           |
|  Just a pothole-patcher on the Information Superhighway.            |
+---------------------------------------------------------------------+



-----------[000441][next][prev][last][first]----------------------------------------------------
Date:      24 May 1994 13:21:10 +0200
From:      rcsacw@rw8.urc.tue.nl (Christ van Willegen)
To:        comp.protocols.tcp-ip
Subject:   Re: Source for WU ftpd available? [Summary]

Hello

Sorry to tell you this is *not* a reply to the source mentioned
above.... I'm just glad you are able to track Voyager (4.6e9 km you
said?)

8 watts of power as far as I can remember.... Great job, guys! Looking
forward to pictures of 25 june(july?) of Jupiter (as you guessed I
meant, didn't you).

Signing off for now...

Christ van Willegen
(account valid 'till october 1st 1994)

-----------[000442][next][prev][last][first]----------------------------------------------------
Date:      24 May 1994 13:25:24 +0200
From:      rcsacw@rw8.urc.tue.nl (Christ van Willegen)
To:        comp.protocols.tcp-ip
Subject:   Sorry about my previous msg...

Sorry all out there....

So, I accidentally hit F(ollowup) instead of R(eply to sender).. Oops. I
used to do this the other way 'round when I started (in an other area).
I used to Reply to the senders instead of Following up, so others could
benefit from me ;-)

Again, sorry for the inconvenience, etc etc
Christ van Willegen
(account STILL valid 'till october 1st 1994)

-----------[000443][next][prev][last][first]----------------------------------------------------
Date:      24 May 1994 16:56:53 GMT
From:      yongzou@cs.sunysb.edu (Yong Zou)
To:        comp.protocols.tcp-ip
Subject:   (help)comunication between DOS and UNIX

Subject: (help)comunication between DOS and UNIX
Newsgroups: comp.protocols.tcp-ip.ibmpc
Summary: 
Keywords: 

Hi, everybody.
Is it posible to build a comunication between DOS and UNIX using sockets
programming?
I tried build a server on DOS (NFS-PC programming toolkit: socket) and a 
client on UNIX. There is a strange behavior:
DOS can write correctly to UNIX; UNIX can't write correctly to DOS.
The code is build on sockets and uses read and write to comunicate. 
Can somebody tell me why?
Thanks.



-----------[000444][next][prev][last][first]----------------------------------------------------
Date:      Tue, 24 May 1994 18:31:16 GMT
From:      simpson@syllabub.tfs.com (Scott Simpson)
To:        comp.protocols.tcp-ip
Subject:   Is there a protocol for encrypting IP packets?

Is there a protocol that will encrypt IP packets on one end and
decrypt them on the other? We currently pay big bucks for a leased
line to England because we are concerned about security. It would be
cheaper for us to get a local connection in England, encrypt packets,
send them through the Internet and decrypt them on the other side. Is
there a protocol that does something like this. I imagine it might be
difficult because the encryption would have to maintain the same size
of the IP packet or cause it to fragment the packet even more.

-----------[000445][next][prev][last][first]----------------------------------------------------
Date:      Tue, 24 May 1994 18:41:59 +0000
From:      vadim@cix.compulink.co.uk (Vadim Lebedev)
To:        comp.protocols.tcp-ip
Subject:   Re: pOMOGITE WOSSTANOWITX DANN


In-Reply-To: <ABPB8tjGO3@bfh.irkutsk.su> denis <denis@bfh.irkutsk.su>

> TITLE: pOMOGITE WOSSTANOWITX DANNYE!!!
> 
> +Message-ID   : <ABPB8tjGO3@bfh.irkutsk.su>
> +From         : denis <denis@bfh.irkutsk.su>
> +Organization : Baikal Stock House, Co Ltd.
> 
>         pOMOGITE WYJTI IZ UVASNOJ SITUACII!
> 
>    dELO W TOM, ^TO PRI RABOTE W NetWare 3.11 NA FAJL-SERWERE EverEX Step-VL
> STALI WOZNIKATX PROBLEMY S SINHRONIZACIEJ DISKOW (PRI OTRAVENII). pRI \TOM
> REZKO UPALA PROIZWODITEELXNOSTX W SETI (IMENNO PO \TOMU MY NE ZAWER[ILI
> RABOTU SBackup'A - KAK NA ZLO!).
>   mY (- @#$%^&) RE[ILI POKA WYKL@^ITX "ZERKALKU" (ZAPUSTILI Install). nO PO
> POLNOMU NEZNANI@ OSOBENNOSTEJ PODLOGO NOWELQ, NE PEREGRUZIW SERWER, GROHNULI
> WS@ INFORMACI@ O TOMAH NA ODNOM DISKE (A ^TO DELATX, ESLI NETWARE RUGALASX,
> ^TO TOMA NA DISKAH ODINAKOWYE?). nU I, ESTESTWENNO TOVE SAMOE OTRAZILOSX I NA
> WTOROJ DISK. w ITOGE MY OSTALISX BEZ ^REZWY^AJNO WAVNYH DANNYH (Q UVE NE
> GOWOR@ O KU^E DOKUMENTACII, BEZWREMENNO PO^IW[EJ S NIMI - BOG S NEJ).
>   bOLX[E NIKAKIH DEJSTWIJ S DISKAMI MY NE PROIZWODILI (UBRALI W STORONKU).
> 
>   i WOT, PRIWYK[IE K WOZMOVNOSTQM ms-dOSA, MY NE MOVEM NE WERITX W WOZMOVNOSTX
> WOSSTANOWLENIQ DANNYH.
>   tak li |to? mOVET ESTX KAKOJ-NIBUDX SOFT? iLI HOTQ BY MOVNO UZNATX, W KAKOM
> MESTE NA WINTE NOWELX DERVIT (-AL) DANNYE O TOMAH?
> 
>   P.S.  nA WSQKIJ SLU^AJ: WINT - Seagate ST3600N.
>                                                      dENIS gOLUBEW
> 
> 
> 

It's funny, the above message is actually a Russian text which i managed 
to decipher:

   Need help in horrible situation!

The problem is that while working on Netware 3.11 on Everex Step-Vl file 
server we've started to experience problems with disk synchronisation 
(with mirroring). At the same time the network performance gone drastically 
down (Because of that we didn't tarminated the SBackup)
We ( intraductible)  decided  to swicth off mirroring for a 
while (launched Install).
But by ignorance we didn't reload the server, and erased all 
volume information
on one of the disks (What we shoul be doing if netware cryed that the volumes
on the disks are same?). So, evedently the same was mirrored to the 
second disk
At the end we've lost extremely important data (and i'm not talking about a 
mountain of the documentation persihed untimely - let it be) 
We didn't try anything more with these disks ( put'em aside).
But accustomed to the psibilities of MS-Dos we can not abanod the belief in
a possiblity to restore the lost data. Is it so? May be there exists some 
software. Or maybe at least is possible to find out where the volume information is (
was) stored on the disk?
 P.S. By the way the disk - Seagate ST4600N.
     Denis  Golubev.




Translation by: 
-----------------------------------------------------------------------
Vadim Lebedev                      |   Kortex International
vadim@cix.compulink.co.uk          |   139-147 av. Paul-Vaillant Couturier
vadim@bix.com                      |   93126 La Courneuve - CEDEX, France



-----------[000446][next][prev][last][first]----------------------------------------------------
Date:      24 May 1994 19:12:12 GMT
From:      emv@garnet.msen.com (Edward Vielmetti)
To:        comp.protocols.tcp-ip
Subject:   Single process TCP based information server

I'd like to get pointers to books and/or available code demonstrating
how to write TCP/IP servers that run as a single, non-forking process,
suitable for serving up lots of queries in a short time without 
having a Unix system spawn a ton of processes.

The continuum I'm going down runs like this:

servers from inetd: easy to write (even shell scripts), 
requires inetd to fork and exec a new process every time,
most inetd's have load limiting code that detect and
destroy "looping" (e.g. busy) servers.  e.g. "gn"

standalone servers: somewhat more work to write; single
long-running process forks other processes to handle queries;
some efficiency gained by keeping inetd out.  e.g. "sgn"

single process server: single, non-forking process handles
all incoming requests, using "select" or some similar queueing
to handle overlapping events; more complicated to get right,
generally requires signals handling or other funky efforts
to do reconfigurations.  e.g. "ntpd", "named".

Any simpified skeleton code, model software, or condensed
"how it's done" information welcomed.

  Edward Vielmetti, vice president for research, Msen Inc. emv@Msen.com
Msen Inc., 320 Miller, Ann Arbor MI  48103 +1 313 998 4562 (fax: 998 4563)
 msen info addresses:   info@msen.com - $20/mo public access Internet
 			occ-info@msen.com - Online Career Center jobs database

-----------[000447][next][prev][last][first]----------------------------------------------------
Date:      24 May 94 19:50:42 GMT
From:      davidm@ittpub.nl
To:        comp.protocols.tcp-ip,comp.protocols.tcp-ip.domains
Subject:   Internet connection w/o Class A or B ?

hi,

Can anyone tell me, how should TCP/IP addressing work where we will have a
TCP/IP network connected to the Internet but where we allocate the address
numbers for our nodes and we use just one Internet address for a gateway ?

How would we telnet or ftp thru' the gateway ?

Is this dependent on the TCP/IP implementation or the Internet service
provider ?

How do users with large networks connect to the Internet without being
registered as Class A or B networks ?

Thanks in advance

------------------------------------------------------------------------
D P Morgan, ITT Publitec R+D BV, Hoekenrode 1, 1102BR Amsterdam, Holland
david@ittpub.nl, ...!sun4nl!ittpub!david, Ph. +31-20-5676737 fax 6910374

-----------[000448][next][prev][last][first]----------------------------------------------------
Date:      24 May 1994 21:24:40 GMT
From:      mjo@iao.ford.com (Mike O'Connor)
To:        comp.protocols.tcp-ip
Subject:   Re: Non-Interactive input to telnet

In article <timc.21.000FA91F@sni.com.au> timc@sni.com.au (Tim Cullen)
writes:

:This question has probably been already addressed in a FAQ.
:
:How can I use telnet with input from a file (on a Unix host) ?
:
:i.e telnet host < cmdlist

It is in the comp.unix.questions FAQ.  I quote:

Subject: How do I run ... interactive programs from a shell script ... ?
Date: Thu Mar 18 17:16:55 EST 1993

3.9)  How do I run 'passwd', 'ftp', 'telnet', 'tip' and other interactive
      programs from a shell script or in the background?

      These programs expect a terminal interface.  Shells makes no
      special provisions to provide one.  Hence, such programs cannot
      be automated in shell scripts.

      The 'expect' program provides a programmable terminal interface
      for automating interaction with such programs.  The following
      expect script is an example of a non-interactive version of
      passwd(1).

        # username is passed as 1st arg, password as 2nd
        set password [index $argv 2]
        spawn passwd [index $argv 1]
        expect "*password:"
        send "$password\r"
        expect "*password:"
        send "$password\r"
        expect eof

      expect can partially automate interaction which is especially
      useful for telnet, rlogin, debuggers or other programs that have
      no built-in command language.  The distribution provides an
      example script to rerun rogue until a good starting configuration
      appears.  Then, control is given back to the user to enjoy the game.

      Fortunately some programs have been written to manage the
      connection to a pseudo-tty so that you can run these sorts of
      programs in a script.

      To get expect, email "send pub/expect/expect.shar.Z" to
      library@cme.nist.gov or anonymous ftp same from
      ftp.cme.nist.gov.

      Another solution is provided by the pty 4.0 program, which runs a
      program under a pseudo-tty session and was posted to
      comp.sources.unix, volume 25.  A pty-based solution using named
      pipes to do the same as the above might look like this:

        #!/bin/sh
        /etc/mknod out.$$ p; exec 2>&1
        ( exec 4<out.$$; rm -f out.$$
        <&4 waitfor 'password:'
            echo "$2"
        <&4 waitfor 'password:'
            echo "$2"
        <&4 cat >/dev/null
        ) | ( pty passwd "$1" >out.$$ )

      Here, 'waitfor' is a simple C program that searches for
      its argument in the input, character by character.

      A simpler pty solution (which has the drawback of not
      synchronizing properly with the passwd program) is

        #!/bin/sh
        ( sleep 5; echo "$2"; sleep 5; echo "$2") | pty passwd "$1"

-- 
 Michael J. O'Connor           |  Internet:  mjo@jobone.srl.ford.com
 Ford Motor Company, OPEO      |  UUCP:      ...!fmsrl7!opeo!mjo
 20000 Rotunda, Bldg. 1-3001   |  Phone:     +1 (313) 248-1260
 Dearborn, MI  48121           |  Fax:       +1 (313) 323-6277

-----------[000449][next][prev][last][first]----------------------------------------------------
Date:      24 May 1994 21:31:23 GMT
From:      mjr@tis.com (Marcus J Ranum)
To:        comp.protocols.tcp-ip
Subject:   Re: Is there a protocol for encrypting IP packets?

[Scott -- First off, fix your mailer. Personal mail to you loops.]

simpson@syllabub.tfs.com (Scott Simpson) writes:
>Is there a protocol that will encrypt IP packets on one end and
>decrypt them on the other? We currently pay big bucks for a leased
>line to England because we are concerned about security. It would be
>cheaper for us to get a local connection in England, encrypt packets,

	You might want to talk to Semaphore technologies, they
have something that does the kind of thing you're looking for.

	Bear in mind that most crypto gear that's "good enough"
for export is "good enough" that someone can crack it if they
care enough. It'll certainly defeat casual scrutiny. There are some
vendors who sell crypto gear made outside the US, which is high
quality.

mjr.

-----------[000450][next][prev][last][first]----------------------------------------------------
Date:      Tue, 24 May 1994 22:07:20 GMT
From:      kbush@harp.aix.calpoly.edu (Kira Kristi Bush)
To:        comp.protocols.snmp,comp.protocols.tcp-ip
Subject:   Interested in tutorial software?


	Sorry if this is the wrong place to post, but I am looking for
people or companies who would be interested in tutorial software
(courses) on TCP/IP, LAN, SNMP, etc.  If you are or know of a company
that might be interested, please let me know or tell me where I might
find that information.  Thanks a lot.

Greg


-----------[000451][next][prev][last][first]----------------------------------------------------
Date:      Tue, 24 May 1994 23:23:50 GMT
From:      jeffx@netcom.com (Jeff Crilly N6ZFX)
To:        comp.protocols.tcp-ip
Subject:   Logging ftpd transfers?

Maybe this is a real easy question...  I haven't been able to find
an answer...

How does one set up an ftp server that logs transfer requests, etc.?
We want to know all events that occur on the server: user connection,
file transfer started, file transfer completed, file transfer failed, user
disconnected, and so on.

Is the solution to get the wu-ftpd from wuarchive and build it?

Is this a standard feature of the berkeley ftpd distribution?

One problem with the wu-ftpd is that I want to run this on
a QNX 4.2 machine which isn't directly supported as one of the ports
in the wu-ftpd archive.  QNX conforms to POSIX, so its possible that
one of the existing ports will just work.  Or that the port isn't
that hard.

I'm trying to get the most bang for the buck: I don't have alot of time,
but I need logging.  Any other suggestions?

Thanks

jeff
jeffx@netcom.com




-----------[000452][next][prev][last][first]----------------------------------------------------
Date:      24 May 1994 23:41:30 GMT
From:      tli@cisco.com (Tony Li)
To:        comp.protocols.tcp-ip
Subject:   Re: IGRP proprietary

In article <2rt0dq$275@ace.mid.net> adar0@routers.com writes:
    
    > Hi, I need to know if the IGRP protocol (CISCO's routing protocol used 
    > in internet) is proprietary of CISCO or other manufacturers handle that.
    
    It is Cisco's proprietary routing protocol.  I do not know of any
    other vendor supporting this protocol, although you might get a
    better answer by reposting this question in /comp/dcom/sys/cisco.

Only folks who are also running our entire operating system, such as DEC.

Tony

-----------[000453][next][prev][last][first]----------------------------------------------------
Date:      Wed, 25 May 1994 00:20:12 GMT
From:      gnn@netcom.com (George Neville-Neil)
To:        comp.protocols.tcp-ip
Subject:   Re: TCP Connection.

vishwamber.yelsangikar.1@nd.edu (Vishwamber Yelsangikar) writes:

>If a physical link is down and there is already a TCP connection what will

The connection will go down, unles the physical link is repaired before
the TCP connection times out.

>happen to that connection?  In my case when our router went down all my
>remotely logged in windows disappeared.  When does TCP connection gets
>disconnected under such circumstances? and how? gracefully or not? 
>Logically there should never be any kind of disconnection.  It should get
>hung during that extended period of router downtime.

It is not clear that "hanging" is a good thing.  If the connection 
hangs how do you a) restart it, b) detect the reason for the hang
etc.


Hope this helps,
George

-- 
gnn@netcom.com

Gentlemen, I will not have you fighting in the War Room.
					--- The President in Dr. Stragelove

-----------[000454][next][prev][last][first]----------------------------------------------------
Date:      Wed, 25 May 1994 01:12:25 GMT
From:      atkinson@itd.itd.nrl.navy.mil (Ran Atkinson)
To:        comp.protocols.tcp-ip
Subject:   Re: Fine-tuning TCP/IP

In article <1994May24.071143.2455@news.datasrv.co.il> logtel@zeus.datasrv.co.il (LOGTEL) writes:

>Is it really possible to somehow fine-tune the packet size for the
>UNIX TCP/IP implementations ?

Not sure what you mean.  The above isn't a well formed question.
Perhaps you would be interested to read the RFC on "Path MTU Discovery" ?

[FAQ: anonymous ftp to ftp.internic.net:rfc/rfc-index will get an
index of all RFCs; all RFCs are online at the same place; please do
NOT ask which number something is, please use the rfc-index online]

Ran
atkinson@itd.nrl.navy.mil



-----------[000455][next][prev][last][first]----------------------------------------------------
Date:      Wed, 25 May 1994 01:14:48 GMT
From:      atkinson@itd.itd.nrl.navy.mil (Ran Atkinson)
To:        comp.protocols.tcp-ip
Subject:   Re: ntp time servers

In article <1994May24.162110.22078@sco.com> leff@sco.com (Bill Leff) writes:
>
>Can someone please email me the address of an ntp time
>server on the internet? 

Wrong newsgroup.  NTP discussions are in comp.protocols.time.ntp, not here.

Answer to question is to read "clock.txt" on louie.udel.edu (anon ftp).

Ran
atkinson@itd.nrl.navy.mil



-----------[000456][next][prev][last][first]----------------------------------------------------
Date:      Wed, 25 May 1994 01:19:08 GMT
From:      atkinson@itd.itd.nrl.navy.mil (Ran Atkinson)
To:        comp.protocols.tcp-ip
Subject:   Re: Is there a protocol for encrypting IP packets?


>simpson@syllabub.tfs.com (Scott Simpson) writes:
>>Is there a protocol that will encrypt IP packets on one end and
>>decrypt them on the other? We currently pay big bucks for a leased
>>line to England because we are concerned about security. It would be
>>cheaper for us to get a local connection in England, encrypt packets,

In article <2rtrnc$qaj@shemesh.tis.com> mjr@tis.com (Marcus J Ranum) writes:

>	You might want to talk to Semaphore technologies, they
>have something that does the kind of thing you're looking for.

Yes.  I believe that Motorola (Scottsdale, Arizona) and Hughes (Southern
California), UUNET Technologies (Falls Church, Virginia), and ANS
(Reston, Virginia) also make products that might be of interest.

>	Bear in mind that most crypto gear that's "good enough"
>for export is "good enough" that someone can crack it if they
>care enough. It'll certainly defeat casual scrutiny. There are some
>vendors who sell crypto gear made outside the US, which is high
>quality.

Imprecisely stated.  DES is exportable to the UK (and other US allied
countries) without much difficulty and is plenty strong for most
folks.


-----------[000457][next][prev][last][first]----------------------------------------------------
Date:      25 May 1994 01:41:40 GMT
From:      mpinones@netmon.mty.itesm.mx (Marco A. Pinones I.)
To:        comp.protocols.tcp-ip.ibmpc,comp.protocols.tcp-ip,comp.protocols.ppp
Subject:   Re: security of SLIP/PPP

Wilfred Mollenvanger (wilfred@zeus.wnc.nedlloyd.nl) wrote:
: : Is your current system secure?  Do you allow dialup access?  Are you
: : connected to the internet?  If you are already connected to the outside
: : world, either by dialup or the internet, adding SLIP or PPP should not
: : make much difference.  It is somewhat more likely to affect the security
: : of the systems which dial in to the net, unless the users are careful with
: : such things as ftp servers.
: You could also use the tcp wrapper program on your sparc to filter
: incoming connections on source address and/or protocol type.
: (cert.sei.cmu.edu:/pub/network_tools/tcp_wrapper.shar I believe)

How do you protect from your terminal server the slip users? Can you
assign multiple passwords to one port or accounts?
Is there any way to accomplish such task? We want to provide our
developers ppp or slip access to the net, but we want also to protect
the access.
Greetings,

-----------[000458][next][prev][last][first]----------------------------------------------------
Date:      Wed, 25 May 94 03:29:39 GMT
From:      ftlofaro@unlv.edu (Frank Lofaro)
To:        comp.protocols.ppp,comp.protocols.tcp-ip
Subject:   Re: NCSA 2.6 and InterSLIP problems

In article <781@moria.macs.sp.unisys.com> bas@moria.macs.sp.unisys.com (Brian Strop) writes:
>
>I have a copy of NCSA Telnet 2.6 and InterSlip 1.0.1.  I am connecting
>to a Unisys ATS II terminal server (which is really a Xylogics Annex
>terminal server under the hood).  I am running at 9.6Kbps and have
>the RFC1044 compression option turned on in my InterSlip configuration.
>I am running an MTU of 1006 bytes.  I am NOT using hardware handshaking (
                                      ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

This is a Very Bad Thing with SLIP! Either you are running without flow 
control at all (lose) or have software flow control (Really Big Lose!).
Software flow control on a line will hose SLIP, since SLIP data contains 
the flow control characters unescaped, and the flow control will eat them, 
freeze up the line, etc. SLIP needs a TOTALLY clean line.



-----------[000459][next][prev][last][first]----------------------------------------------------
Date:      Wed, 25 May 94 05:09:11 GMT
From:      hal9001@panix.com (Robert A. Rosenberg)
To:        comp.protocols.tcp-ip
Subject:   Re: MacTCP & SLIP Help!!

In Article <2rqc4e$g3m@sheckley.cnam.fr>, bortz@cnam.cnam.fr (Stephane
Bortzmeyer) wrote:
>May be in theory. But in practice (which was the concern of the original
>poster), MacPPP (*one* specific implementation) is less buggy than any
>*Macintosh* SLIP implementation.

Thank you for coming to my defense on recommending that the user look into
MacPPP  _since they are have problems with Interslip_.

I'll throw out another neat feature of MacPPP that InterSLIP (and probably
other SLIPs) does not have. A user on my provider incorrectly set up his
MacTCP to use SERVER addressing (because the _docs said to_) contrary to the
"Welcome to Panix.COM - Here is how to set MacTCP for your SLIP/PPP
Connection" Message. The user could not connect (there is not IP Server on
the line - all SLIP/PPP connects are Dedicated IP Address). Since the
Provider KNEW the IP address (from the Logon) it offered to negotiate the IP
address (as opposed to just supplying one as in the case of a Server) and
MacPPP (which has IP Address Negotiation code) accepted the "suggestion" and
worked as if MacTCP had (correctly) been set to Manually.

-----------[000460][next][prev][last][first]----------------------------------------------------
Date:      25 May 1994 05:52:19 GMT
From:      iotov@gluttony.ugcs.caltech.edu (Mihail Stilianov Iotov)
To:        comp.protocols.tcp-ip,comp.protocols.tcp-ip.domains
Subject:   Re: Internet connection w/o Class A or B ?

davidm@ittpub.nl writes:

>Can anyone tell me, how should TCP/IP addressing work where we will have a
>TCP/IP network connected to the Internet but where we allocate the address
>numbers for our nodes and we use just one Internet address for a gateway ?
 
>How would we telnet or ftp thru' the gateway ?

This might be possible but I wouldn't advise. You might get the gateway to
route IP packages to the internet from your other machines. But consider 
what happens if you happened to choose an IP address for your local machines
that already exists. The return packets from the host your are trying to
ftp to get routed to the real host with the official address not to the machine
that initiated ftp ! If you came up with address that no one uses it might
be able to advertise it to the whole internet by RIP, but will work only
if there are dynamic routers willing to listen and until someone gets assigned
the address. 

So my advice is : telnet to the gateway and ftp from there. Then ftp back to
the local machine. Same with telnet.

-----------[000461][next][prev][last][first]----------------------------------------------------
Date:      25 May 94 15:11:10 -0500
From:      harvey@indyvax.iupui.edu (James Harvey)
To:        comp.protocols.tcp-ip
Subject:   Re: Assigning IP addresses to a CISCO WAN

In article <2rsufi$207@ace.mid.net>, adar0@routers.com writes:
>
>> We currently
>> are having a discussion about how to assign IP numbers to the WAN
>> Synch ports.
>>
>> Each LAN has its own class C IP subnet, and currently the two
 synchronous
>> ports comprise a third class C subnet.
>>
> Mark,
> Technically, either solution will work and will not cause any long
> term problems.  However, business growth and/or practicality may
> impact which approach is more appropriate.  From a business perspective,
> if the business grows, using a Class-C for the backbone net has the
> advantage of being able to be "managed" by a separate group while
> the end Class-C's are managed by their respective groups.  The
> disadvantage is that you cannot break ONE class-c address space into
> two geographic areas and expect the sessions to function through a
> second class-c address space.  All addresses within a class must be
> contiguous.
> From a practicality perspective, assuming the business does not grow
> much, burning a full class-c for a couple of serial links is quite
> wastefull.

The original article has expired here so please excuse me if this has
already been pointed out.

Ciscos have an "IP unnumbered" option that enables IP processing on a
serial interface without having to assign an IP address to the interface.
You have to specify another (non-unnumbered) interface on the router to be
used to get a source address for packets generated by the interface (for
example, for a routing update) and there are a few restrictions: HDLC only
(no X.25), can't ping (but SNMP works) and no IP security options.  But it
does save you from having to waste subnets on serial links.
--
James Harvey   harvey@iupui.edu   IUPUI OIT Networks and Systems Tech Support
Disclaimer:  These are my own opinions.  I do not speak for Indiana University.

-----------[000462][next][prev][last][first]----------------------------------------------------
Date:      Wed, 25 May 1994 09:29:31 GMT
From:      csgoh@iti.gov.sg (Goh Cheng-Seng)
To:        comp.protocols.tcp-ip
Subject:   public domain DHCP server?

Hi,

Can anyone tell me whether there exists public domain Dynamic Host
Configuration Protocol (DHCP) server software for SunOS 4.1.x? DHCP is
supposed to support dynamic IP address allocation. 

Thanks for your attention.


-- 
Goh Cheng-Seng                |  Computer Information Systems Department
Email: csgoh@ncb.gov.sg       |  National Computer Board, Singapore 
Phone: (65)-772-0451          |  71 Science Park Drive
Fax  : (65)-778-9641          |  Singapore 0511

-----------[000463][next][prev][last][first]----------------------------------------------------
Date:      25 May 1994 10:03:56 GMT
From:      kjetil@newton.spd.eee.strathclyde.ac.uk (Kjetil Rossavik)
To:        comp.protocols.tcp-ip
Subject:   Protocol Execution Overhead

Hi,

I am looking at the responsiveness of applications running on a network,
and I need some indications to the execution time of protocol layers.

Does anybody have any figures for TCP and IP, connection set-up, data
transfer (short packets, medium, long packets), preferrably broken 
down on protocol layer.

Please indicate system (Sun sparc preferred).

Thanks,
Kjetil.

PS. I am also looking for corresponding figures for X.25 and HDLC.

-----------[000464][next][prev][last][first]----------------------------------------------------
Date:      25 May 1994 10:46:17 GMT
From:      jim@cs.strath.ac.uk (Jim Reid)
To:        comp.protocols.tcp-ip
Subject:   Re: ntp time servers

In article <1994May24.162110.22078@sco.com> leff@sco.com (Bill Leff) writes:

   Can someone please email me the address of an ntp time
   server on the internet? 

A list of NTP servers and their access policies can be picked up by
anonymous ftp from louie.udel.edu: ntp/clock.txt

-----------[000465][next][prev][last][first]----------------------------------------------------
Date:      25 May 94 09:56:54 +0100
From:      kloska@mpimg-berlin-dahlem.mpg.de
To:        comp.protocols.tcp-ip
Subject:   [Q] Appletalk via TCP/IP

Hello,

I'm using MacTCP via MacPPP to connect to the ultrix system (V4.2) in our
office, and everything so far is just fine with running ftp, telnet etc.

But...

All the outher Mac's connected directly via Ethernet are able to use (1)
the printers and (2) public servers, and that makes me kind of gready. So
my sysob told that all this kind of fancy stuff would be possible, if there's
a program on my mac site that encapsulates Appletalk into TCP/IP and a partner
on the ultrix site which depackes these and sends them to the
apropriate server. 

Does this kind of solution exist ?

I've already checked out the CAP and ADATALK programs but they seem to rely
on IP encapsulation.

Thanks....


Sebastian Kloska (kloska@mpimg-berlin-dahlem.mpg.de)

-----------[000466][next][prev][last][first]----------------------------------------------------
Date:      Wed, 25 May 1994 12:21:07 GMT
From:      w-rolph@ds.mc.ti.com (Don Rolph)
To:        comp.protocols.tcp-ip
Subject:   Re: Two NIC in the same ethernet segment, possible?

In article <2rrle2$ltt@nuscc.nus.sg> ccetc@nusunix1.nus.sg (SuperFast) writes:
>From: ccetc@nusunix1.nus.sg (SuperFast)
>Subject: Two NIC in the same ethernet segment, possible?
>Date: 24 May 1994 01:31:46 GMT


>Is it possible to put two Ethernet cards in a UNIX box, and assign them
>two different IP address for the same subnet?
 
>The intension is to split network traffics/load between the two cards so that
>higher performance can be achived. The other possibility is to put a UNIX server
>machine with two NICs to a EtherSwitch so that the total bandwidth
>become 2x10Mbps.
 
>Is this possible at all?
 
>--
>ccetc@leonis.nus.sg
>National University of S'pore

It is certainly possible to install two nic and have outside nodes connect to 
either of them by name or address (note there is no automatic spill over - you 
will name each interface, for example, node1 and node2 and remote users will 
connect to either node1 or node2 as they choose).  NOte however, based on some 
painful lessons in our shop, if the system is running a Berkeley derivative 
UNIX (thats where our unfortunate test was run), the system will choose 
adapter one to route out all its traffic (after all adapter 1 does have access 
to the desired subnet).

If you are using filtering bridges this can be very very bad.  It also doesn't 
load share between the boards.

In short - use two adpater for two different subnets (even if on the same 
wire), but dont try to load split.

Regards.
 
Don Rolph w-rolph@ds.mc.ti.com WD3 MS10-13 (508)-699-1263

-----------[000467][next][prev][last][first]----------------------------------------------------
Date:      Wed, 25 May 1994 12:23:30 GMT
From:      w-rolph@ds.mc.ti.com (Don Rolph)
To:        comp.protocols.tcp-ip
Subject:   Re: (help)comunication between DOS and UNIX

In article <2rtbkl$j7b@newsserv.cs.sunysb.edu> yongzou@cs.sunysb.edu (Yong Zou) writes:
>From: yongzou@cs.sunysb.edu (Yong Zou)
>Subject: (help)comunication between DOS and UNIX
>Date: 24 May 1994 16:56:53 GMT
 
>Subject: (help)comunication between DOS and UNIX
>Newsgroups: comp.protocols.tcp-ip.ibmpc
>Summary: 
>Keywords: 
 
>Hi, everybody.
>Is it posible to build a comunication between DOS and UNIX using sockets
>programming?
>I tried build a server on DOS (NFS-PC programming toolkit: socket) and a 
>client on UNIX. There is a strange behavior:
>DOS can write correctly to UNIX; UNIX can't write correctly to DOS.
>The code is build on sockets and uses read and write to comunicate. 
>Can somebody tell me why?
>Thanks.

It can certainly be done - I have done it using the sample code from the 
microsofot lanman tcp/ip socket development kit.  Sounds like you have a bug 
in your code.



Regards.
 
Don Rolph w-rolph@ds.mc.ti.com WD3 MS10-13 (508)-699-1263

-----------[000468][next][prev][last][first]----------------------------------------------------
Date:      25 May 94 18:30:42
From:      vixie@vix.com (Paul A Vixie)
To:        comp.protocols.tcp-ip,comp.protocols.tcp-ip.domains
Subject:   Re: Internet connection w/o Class A or B ?

> Can anyone tell me, how should TCP/IP addressing work where we will have a
> TCP/IP network connected to the Internet but where we allocate the address
> numbers for our nodes and we use just one Internet address for a gateway ?
>
> How would we telnet or ftp thru' the gateway ?

You can get the firewall toolkit from tis.com:~ftp to get good proxy ftp
and proxy telnet daemons.  This will save you from having to allocate user
accounts on your gateway machine.

> Is this dependent on the TCP/IP implementation or the Internet service
> provider ?

No.  Not at all.

> How do users with large networks connect to the Internet without being
> registered as Class A or B networks ?

At least one.  It's becoming more common.  See RFC 1597, Address Allocation
for Private Internets.
--
Paul Vixie
Redwood City, CA
decwrl!vixie!paul
<paul@vix.com>

-----------[000469][next][prev][last][first]----------------------------------------------------
Date:      25 May 1994 14:57:29 GMT
From:      mgb722c@rs730.gsfc.nasa.gov
To:        comp.protocols.tcp-ip
Subject:   Gopher Server on AIX

Hey all...

I am attempting to create a Gopher server on an IBM RS/6000 box
to serve about 50 users. I have downloaded the Gopher server code
from the University of Minnesota, and it does include information
on how to negotiate via the gopher *client*. However, it gives
minimal information on how to set up menus, links, and the like
necessary for a gopher *server*. Where can I get this info? Is 
there an RFC on this somewhere? If anyone has idea where to get
this info, please post or e-mail me.

Thanks in Advance,

/\\yron

*************************************************
*  Myron Bradshaw - Facility Manager                            *
*  Curtis Management Company, Inc.                               *
*  NASA/Goddard Space Flight Center                             *
*  Greenbelt, MD 20771                                                 *
*  E-Mail: mgb@genie.gsfc.nasa.gov, or                            *
*            mgb722c@rs730.gsfc.nasa.gov                           *
*************************************************
*Jesus not only saves, He also frequently makes Backups.*
*************************************************
*      Get confident, stupid!!!!                                         *
*************************************************


-----------[000470][next][prev][last][first]----------------------------------------------------
Date:      25 May 94 15:43:15 GMT
From:      nieves@eucmvx.sim.ucm.es
To:        comp.protocols.tcp-ip
Subject:   Looking for LPR and LPD

We have installed the Winsock Print Manager and Print Spooler (WSLPP-11)
for PC and have LPD server UCX 2.0E in a VMS machine, and we have some 
problems when print from VMS to PC.
With some spanish characters (enya "ñ", vowels with accent, etc) the program 
doesn't print its correctly, an often some times are truncated.
We have tried other software for PC, LPD15, but it isn't a resident program
and doesn't work with Microsoft Windows.
Anybody knows other LPD for PC , public software that can solve our problems
and where is it?
And other LPD server software for VMS (LPD UCX software you have to define 
a queue for each PC) that you can print to remote IP address?

-----------[000471][next][prev][last][first]----------------------------------------------------
Date:      25 May 94 15:53:20 GMT
From:      mbt@ESD.3Com.COM (Brad Turner)
To:        comp.protocols.tcp-ip
Subject:   Re: Assigning IP addresses to a CISCO WAN

adar0@routers.com writes:
>> We currently
>> are having a discussion about how to assign IP numbers to the WAN
>> Synch ports.
>> 
>> Each LAN has its own class C IP subnet, and currently the two 
 synchronous
>> ports comprise a third class C subnet. 
>> 
>Mark,
>Technically, either solution will work and will not cause any long
>term problems.  However, business growth and/or practicality may
>impact which approach is more appropriate.  From a business perspective,
>if the business grows, using a Class-C for the backbone net has the
>advantage of being able to be "managed" by a separate group while 
>the end Class-C's are managed by their respective groups.  The
>disadvantage is that you cannot break ONE class-c address space into
>two geographic areas and expect the sessions to function through a
>second class-c address space.  All addresses within a class must be
>contiguous.
>From a practicality perspective, assuming the business does not grow
>much, burning a full class-c for a couple of serial links is quite
>wastefull.

If you are running OSPF there is no need to "burn" a full class C
subnet, make use of variable length subnet masks and mask the WAN
links with 30 bits (ie 255.255.255.252)

-brad-
--
v^v^v^v^v^v^v^v^v^v^v^v^v^v^v^v^v^v^v^v^v^v^v^v^v^v^v^v^v^v^v^v^v^v^v^v^v^v^v^v
Brad Turner | 5400 Bayfront Plaza  | Marketing Engineer   | (408) 764-5261
3Com Corp.  | Santa Clara CA, 95052| mbt@NSD.3Com.Com     | (408) 764-5002 fax
                     A proper eDUCATIon is the only answer.

-----------[000472][next][prev][last][first]----------------------------------------------------
Date:      Wed, 25 May 1994 16:07:48 GMT
From:      xxmcleis@indsvax1.indstate.edu
To:        comp.protocols.tcp-ip
Subject:   Unix computer as a router?

Can this be done? Can anyone give my an example of a computer acting
as a router, specifically to an Internet provider w/ 56k digital line?

I realize there may be a performance limitation, but we may not have the
money to get a *real* router like a CISCO. 

Thanks,
Mike

-----------[000473][next][prev][last][first]----------------------------------------------------
Date:      25 May 1994 16:33:50 GMT
From:      barmar@think.com (Barry Margolin)
To:        comp.protocols.tcp-ip
Subject:   Re: Non-Interactive input to telnet

In article <timc.21.000FA91F@sni.com.au> timc@sni.com.au (Tim Cullen) writes:
>This question has probably been already addressed in a FAQ.

I don't think it's in the tcp-ip FAQ yet (it only recently got started, and
is still pretty minimal).  I've appended the relevant section of the Unix
FAQ below.u

>How can I use telnet with input from a file (on a Unix host) ?
>
>i.e telnet host < cmdlist

The problem with this is that telnet exits as soon as it gets an EOF on its
input.  So it just slurps up the entire file, sends it to the remote host,
and then exits without waiting for the output.

Here's the Unix FAQ entry:

Subject: How do I run ... interactive programs from a shell script ... ?
Date: Thu Mar 18 17:16:55 EST 1993

3.9)  How do I run 'passwd', 'ftp', 'telnet', 'tip' and other interactive
      programs from a shell script or in the background?

      These programs expect a terminal interface.  Shells makes no
      special provisions to provide one.  Hence, such programs cannot
      be automated in shell scripts.

      The 'expect' program provides a programmable terminal interface
      for automating interaction with such programs.  The following
      expect script is an example of a non-interactive version of
      passwd(1).

	# username is passed as 1st arg, password as 2nd
	set password [index $argv 2]
	spawn passwd [index $argv 1]
	expect "*password:"
	send "$password\r"
	expect "*password:"
	send "$password\r"
	expect eof

      expect can partially automate interaction which is especially
      useful for telnet, rlogin, debuggers or other programs that have
      no built-in command language.  The distribution provides an
      example script to rerun rogue until a good starting configuration
      appears.  Then, control is given back to the user to enjoy the game.

      Fortunately some programs have been written to manage the
      connection to a pseudo-tty so that you can run these sorts of
      programs in a script.

      To get expect, email "send pub/expect/expect.shar.Z" to
      library@cme.nist.gov or anonymous ftp same from
      ftp.cme.nist.gov.

      Another solution is provided by the pty 4.0 program, which runs a
      program under a pseudo-tty session and was posted to
      comp.sources.unix, volume 25.  A pty-based solution using named
      pipes to do the same as the above might look like this:

	#!/bin/sh
	/etc/mknod out.$$ p; exec 2>&1
	( exec 4<out.$$; rm -f out.$$
	<&4 waitfor 'password:'
	    echo "$2"
	<&4 waitfor 'password:'
	    echo "$2"
	<&4 cat >/dev/null
	) | ( pty passwd "$1" >out.$$ )

      Here, 'waitfor' is a simple C program that searches for
      its argument in the input, character by character.

      A simpler pty solution (which has the drawback of not
      synchronizing properly with the passwd program) is

	#!/bin/sh
	( sleep 5; echo "$2"; sleep 5; echo "$2") | pty passwd "$1"
-- 
Barry Margolin
System Manager, Thinking Machines Corp.

barmar@think.com          {uunet,harvard}!think!barmar

-----------[000474][next][prev][last][first]----------------------------------------------------
Date:      Wed, 25 May 94 17:06:41 GMT
From:      Paul_Lahaie@achilles.net (Paul JY Lahaie)
To:        comp.protocols.tcp-ip
Subject:   Re: Two NIC in the same ethernet segment, possible?

In article <w-rolph.442.00075A4E@ds.mc.ti.com> w-rolph@ds.mc.ti.com (Don Rolph) writes:
>In short - use two adpater for two different subnets (even if on the same 
>wire), but dont try to load split.

    I was under the impression that Ethernet send all the packets to all the 
stations.  Therefore wouldn't 2 Ethernets on the same wire have a max 
throughput of 10Mbps?

                                               - Paul
--
Paul JY Lahaie                                            pjlahaie@achilles.net


-----------[000475][next][prev][last][first]----------------------------------------------------
Date:      Wed, 25 May 1994 17:23:23 GMT
From:      sybesma@unirsvl.rsvl.unisys.com (Eric Sybesma)
To:        comp.protocols.tcp-ip,comp.protocols.tcp-ip.domains
Subject:   Re: Internet connection w/o Class A or B ?

In <2rup2j$qiu@gap.cco.caltech.edu> iotov@gluttony.ugcs.caltech.edu (Mihail Stilianov Iotov) writes:

>This might be possible but I wouldn't advise. You might get the gateway to
>route IP packages to the internet from your other machines. But consider 
>what happens if you happened to choose an IP address for your local machines
>that already exists. The return packets from the host your are trying to
>ftp to get routed to the real host with the official address not to the machine
>that initiated ftp ! If you came up with address that no one uses it might
>be able to advertise it to the whole internet by RIP, but will work only
>if there are dynamic routers willing to listen and until someone gets assigned
>the address. 
 
>So my advice is : telnet to the gateway and ftp from there. Then ftp back to
>the local machine. Same with telnet.

Just to point out that there is no reason that this process couldn't be
automated.   That is, run a client/server type of application from the
local network machines to the internet machine that knows which hosts
are ftping what and can make the transaction transparent to the users 
on the local hosts.  Or connect with telnet or another similar utility
and automatically start another connection to an internet machine.  If
the use is minimum, the one internet machine could handle this sort of
"responsibility".  
-- 
Eric Sybesma, Unisys     | sybesma@unirsvl.rsvl.unisys.com | (612) 635-3380 |
Site Management Products | "On the information super-highway, I'm a BMW."   | 
Roseville, MN 55113      | self.am = 0;if (self.mode == thinking) self.am++;| 

-----------[000476][next][prev][last][first]----------------------------------------------------
Date:      25 May 1994 17:25:35 GMT
From:      danjackson@vnet.ibm.com (Dan Jackson)
To:        comp.protocols.tcp-ip
Subject:   Re: Connecting PCs & wkstations...

In <2rsvbj$207@ace.mid.net>, adar0@routers.com writes:
>
>There are slight differences between the two and
>you probably won't see it unless you use a sniffer (or some other
>diagnostic tracing tool) to look at the individual packets.  The
>problem relates to two items:  1) when IBM popularized token ring,
>their design shifted the hardware address onto the wire in reverse
>bit order from ethernet standards, and, 2) some software implementations
>use a couple of bits in the hardware address to mean special things
>like group addressing, user administered addressing and multicasting.
>

In case you're interested, the algorithm for swapping between
Ethernet and Token Ring addresses follows.  Note that the algorithm works
whether you're converting from Ethernet to Token Ring or visa versa.

The algorithm:
  (Assumes converting from Ethernet to Token Ring to ease wordage).
  1. Write out the Ethernet address (all 6 bytes) in hexadecimal.
  2. Within each byte swap the hexadecimal digits.
  3. Now within each hexadecimal digit reverse the bit order.

  Example:  Convert  Ethernet 0200 536B E000  to Token Ring format

                 1) Start w/  0200 536B E000
                 2) gets you  2000 35B6 0E00
                 3) gets you  4000 CAD6 0700  == your Token Ring address


Dan Jackson    danjackson@vnet.ibm.com
IBM DOS TCP/IP Development


-----------[000477][next][prev][last][first]----------------------------------------------------
Date:      25 May 94 19:02:51 GMT
From:      poorman@convex.com (Peter W. Poorman)
To:        comp.protocols.tcp-ip
Subject:   Re: Socket read question

A read on a socket blocks until one or more bytes of data arrives from the
other end.  It then gives you those bytes.  This may be fewer than the 
number of bytes passed to write() on the other end.

This is a natural consequence of the byte-stream nature of socket connections.
The O.S. can't tell where one message ends and another begins -- that's totally
up to the applications.

You must repeatedly call read() until you get all the bytes you expect, or
the read() returns < 1.

-- Pete
   poorman@convex.com

-----------[000478][next][prev][last][first]----------------------------------------------------
Date:      26 May 1994 09:12:38 -0700
From:      lstowell@pyrnova.mis.pyramid.com (Lon Stowell)
To:        comp.protocols.tcp-ip
Subject:   Re: Two NIC in the same ethernet segment, possible?

In article <2rrle2$ltt@nuscc.nus.sg> ccetc@nusunix1.nus.sg (SuperFast) writes:
>
>Is it possible to put two Ethernet cards in a UNIX box, and assign them
>two different IP address for the same subnet?

  This is HIGHLY dependent on the implementation of the IP on the
  specific box.  For example, on some Lachman based SVR4 systems you
  may discover that packets are coming IN on both interfaces but
  are only being sent on ONE of the two...and if it fails the box is
  essentially dead.  

  I honestly wouldn't recommend it unless you know the internals of
  your Unix box well.  

  See the "DISCUSSION" on "Strong ES Model" vs "Weak ES Model" in 
  "3.3.4.2 Multihoming Requirements" in RFC 1122 for gory details.



-----------[000479][next][prev][last][first]----------------------------------------------------
Date:      Wed, 25 May 1994 22:46:04 GMT
From:      heberlei@cs.ucdavis.edu  (Louis Todd Heberlein)
To:        comp.protocols.tcp-ip
Subject:   server cannot bind() again, help?

I have set up a single-process, concurrent server with TCP
sockets, and I am having a problem doing bind() on the second
run of the server; I get an address already in use error.

I have the following situation:
    if the server wants to quit (for whatever reason), I call close() on
    all sockets, including the socket which is doing listen() for the server.
    This causes all the clients to shutdown (also calling close() on all
    their sockets).

    However, if I try to run the server again, listening
    to the same port, I get the bind() error. (If I wait a few minutes,
    I can run the server again and the port will be available.)


On the other hand,
    if all the clients quit first (closing their sockets), and then
    the server finishes (closing its sockets), it can call bind()
    on a run immediately after.

It appears that a single-process, concurrent server over TCP must get
the clients to close() their sockets first before the server does; otherwise,
the server will not be able to immediately run again listening to the same
port.

Can I overcome this?  (I have tried shutdown(), but w/o success).

Thanks,

Todd Heberlein					heberlei@cs.ucdavis.edu

-----------[000480][next][prev][last][first]----------------------------------------------------
Date:      Wed, 25 May 1994 22:53:29 GMT
From:      okuyama@netcom.com (Darin Okuyama)
To:        comp.protocols.tcp-ip,comp.sys.sun.admin,comp.sys.sun.misc
Subject:   Workstation as router

I understand that one could use a workstation as a router.  I
have the following questions:

  1. What are the disadvantages of using a workstation as
     a router?  What can a router do that a workstation
     cannot do?  What can a router do better than a work-
     station?

  2. What are the advantages of using a workstation as a
     router?  What can a workstation do that a router can-
     not do?  What can a workstation do better?

  3. How powerful must the workstation be to effectively
     handle a 100 node network (moderately heavy traffic)?
     Answer in terms of Sun workstations if possible.

  4. How does one configure a UNIX workstation to be a
     router?

Thank you.

Darin Okuyama


-----------[000481][next][prev][last][first]----------------------------------------------------
Date:      Wed, 25 May 1994 23:22:18 GMT
From:      ctc@netcom.com (ARTHUR HSIAO)
To:        comp.protocols.tcp-ip
Subject:   TCP & UDP

I am new to this area, but I have a small question on TCP and UDP.  From what
I learned, TCP provides a connection-oriented service, which is like a phone 
call, and consequently when the connection is set up, the client doesn't need 
to address every message; it simply sends the data over the link.

UDP, however, provides a connectionless service, which is similar to sending 
a message by the post office.  Therefore, the client has to address each 
message before it's passed to the transport provider.

To my understanding, there is no such distinction between TCP and UDP.  By
looking at their headers, I can only feel TCP will guarantee messages would 
arrive at the destination in the same order they were sent.  UDP doesn't have
such guarantee.

Please correct me if I am wrong.

--
Arthur
ctc@netcom.com

-----------[000482][next][prev][last][first]----------------------------------------------------
Date:      26 May 1994 12:28:54 -0700
From:      lstowell@pyrnova.mis.pyramid.com (Lon Stowell)
To:        comp.protocols.tcp-ip
Subject:   Re: Unix computer as a router?

In article <CqD8t1.47D@onyx.indstate.edu> xxmcleis@indsvax1.indstate.edu writes:
>
>Can this be done? Can anyone give my an example of a computer acting
>as a router, specifically to an Internet provider w/ 56k digital line?

   Yes....if you have all the software you need.  Most unix hosts
   will act as routers.  Newer ones tend to turn IPFORWARDING off as
   the default for security reasons, but you just turn it back on and
   use routed or manually configured route add statements.

   AS far as the 56 kb link, the normal practice is to run IP over
   X.25 on that link.  How well this works depends on what is on the
   other end of the link, and whether or not your unix host has the
   software and protocols to communicate with the black box at the far
   end.

   If you repost with your unix O/S model and rev level, and what is
   on the other end of the 56 Kb link, very likely someone can give
   you specific config file and package names.
>
>I realize there may be a performance limitation, but we may not have the
>money to get a *real* router like a CISCO. 
>
   Works pretty good, but as a router, most unix boxes are good unix
   boxes....that's why Cisco stock is so high these days. >:-)



-----------[000483][next][prev][last][first]----------------------------------------------------
Date:      26 May 1994 09:33:05 -0400
From:      alfredh806@aol.com (AlfredH806)
To:        comp.protocols.tcp-ip
Subject:   WINQVTNET Users

We are working with WNQVTNET version 3.96 and are enjoying the
product but are having a
problem in the area of printing.

We wish to send everything via transparent print because we are
formatting all of the data
in the billing system prior to printing using ESC sequences. 
However, when we do this, all of the
output goes into a ~xxxxxxx.tmp file in the Windows directory and the
job never makes it to the
printer.

When using print manager the print manager icon pops up on the bottom
of the screen but when you open it, there are no jobs in the queue.

We are a little lost at this point and would appreciate any
assistance WINQVTNET users may provide.

Sincerely,



Alfred S. Harding
ComputerLand Medical Consulting

-----------[000484][next][prev][last][first]----------------------------------------------------
Date:      26 May 94 03:05:42 GMT
From:      visser@convex.com (Lance Visser)
To:        comp.protocols.tcp-ip
Subject:   Re: TCP Connection.

In <vishwamber.yelsangikar.1-240594113314@vish.cc.nd.edu> vishwamber.yelsangikar.1@nd.edu (Vishwamber Yelsangikar) writes:

+>If a physical link is down and there is already a TCP connection what will
+>happen to that connection?  In my case when our router went down all my
+>remotely logged in windows disappeared.  When does TCP connection gets
+>disconnected under such circumstances? and how? gracefully or not? 
+>Logically there should never be any kind of disconnection.  It should get
+>hung during that extended period of router downtime.

	There is a socket option for tcp called keepalive.  When keepalive
is turned on for a particular socket, inactivity can cause the connection
to time out and explicitly fail. 

	Without keepalive, I think the tcp connection should hang around
"forever". IF the only problem is an intermediate link going down.
if the remote end of the connection has crashed, when that remote system comes 
back up any packets from old tcp connections (from before the crash)
will generate a tcp reset response which will kill the connection.
	Certain applications sliently set keepalive for you because they
think you want a timeout rather than perhaps persisting forever.





-----------[000485][next][prev][last][first]----------------------------------------------------
Date:      26 May 1994 11:35:26 -0400
From:      gmichaud@gandalf.ca (Guy Michaud)
To:        comp.protocols.tcp-ip
Subject:   Re: Trumpet Winsock

jim.mcneill@almac.co.uk (Jim Mcneill) writes:

>Anyone tell me where I can get a hand on this Trumper Winsock thingie?
 
>Jim
>---
> * 1st 1.11 #324 * 

Try archive.orst.edu, or ftp.cica.indiana.edu.  Look for wtpkt10a.zip


-----------[000486][next][prev][last][first]----------------------------------------------------
Date:      26 May 1994 11:39:47 -0400
From:      gmichaud@gandalf.ca (Guy Michaud)
To:        comp.protocols.tcp-ip
Subject:   Re: Two NIC in the same ethernet segment, possible?

ccetc@nusunix1.nus.sg (SuperFast) writes:


>Is it possible to put two Ethernet cards in a UNIX box, and assign them
>two different IP address for the same subnet?
 
>The intension is to split network traffics/load between the two cards so that
>higher performance can be achived. The other possibility is to put a UNIX server
>machine with two NICs to a EtherSwitch so that the total bandwidth
>become 2x10Mbps.
 
>Is this possible at all?

Sure.  But you`re running into the possibility of having only one ethernet active at a time; unless the IP network address is different.  Then, routd will 
perform routing between the segments.

--
Guy


-----------[000487][next][prev][last][first]----------------------------------------------------
Date:      Thu, 26 May 1994 04:59:52 GMT
From:      kga@lmsc.lockheed.com (Ken Andrews)
To:        comp.protocols.tcp-ip
Subject:   BIND

Hello,

I am looking for the source code for the latest version of BIND
for our DNS servers. Can anyone point me to an anonymous ftp site?

Appreciate any help...

Ken Andrews

-----------[000488][next][prev][last][first]----------------------------------------------------
Date:      26 May 1994 11:41:18 -0400
From:      gmichaud@gandalf.ca (Guy Michaud)
To:        comp.protocols.tcp-ip
Subject:   Re: Fine-tuning TCP/IP

logtel@zeus.datasrv.co.il (LOGTEL) writes:

>Hi, TCP/IP gurus
 
>Is it really possible to somehow fine-tune the packet size for the
>UNIX TCP/IP implementations ?

Certainly.  But, please give more info; which Unix, which TCP/IP

--
Guy


-----------[000489][next][prev][last][first]----------------------------------------------------
Date:      26 May 1994 17:02:00 -0700
From:      hvora@hsc.usc.edu (Heena Vora)
To:        comp.dcom.servers,comp.protocols.tcp-ip
Subject:   Routing questions on terminal servers and SLIP/PPP

Hi all:

I'm trying to get some info on terminal servers and slip/ppp routing.  I'm 
looking to get a server that dynamically assigns IP addresses. 

If we got 2 (or more) of such servers, and hooked them to a common
modem pool, a user calling in could be connected to one or another 
at any time (obviously)

So in my routing tables (on my host), how do I tell it that it needs to 
look in 2 gateways (for the 2 servers) to find the SLIP/PPP dialer?  

This may be elementary, but I'm just getting a crash course in tcpip.

Thanks
Heena


-----------[000490][next][prev][last][first]----------------------------------------------------
Date:      26 May 1994 15:05:12 -0500
From:      yns1@Ra.MsState.Edu (Yeak Nai Siew)
To:        comp.sys.mac.comm,comp.protocols.tcp-ip
Subject:   Re: Q: NCSA Tel 2.6 - TCP Error netpush -23008

In article <marilyn-260594133845@monroe.ocis.temple.edu>,
Goddess of Macgic! <marilyn@monroe.ocis.temple.edu> wrote:
>
>TCP Error:  failed to get TCP Status.
>netpush -23008
>
>Could this be it due to the fact that I am running MacHTTP 1.3 and while
>somebody was accessing my homepage, NCSA Tel 2.6 didn't get enough TCP 
>status to send/recieve packets.
>
I got netpush error but no TCP error. What I did is to have NCSA Telnet 2.6
running with Server Mode turn on (with or without passwd). Then I ftp from
another computer to it using Fetch (or any ftp program). If during getting
file process I cancel the transmission, the netpush error will be shown in
the Mac that run NCSA Telnet. Sometime it crashes, sometime not.

Any idea?



===================== Apple Macintosh PowerBook user =======================
-Yeak                                                         <<PPP active>>
yns1@Ra.MsState.Edu				     "Nothing Is Impossible"

-----------[000491][next][prev][last][first]----------------------------------------------------
Date:      Thu, 26 May 1994 13:38:45 -0400
From:      marilyn@monroe.ocis.temple.edu (Goddess of Macgic!)
To:        comp.sys.mac.comm,comp.protocols.tcp-ip
Subject:   Q: NCSA Tel 2.6 - TCP Error netpush -23008

Greets-

First of all many thanks to J. Browne for crafting such a fine tcp based
telnet protocol session for the mac, use it all the time.

This was the first time I encountered an/this error.

TCP Error:  failed to get TCP Status.
netpush -23008

Could this be it due to the fact that I am running MacHTTP 1.3 and while
somebody was accessing my homepage, NCSA Tel 2.6 didn't get enough TCP 
status to send/recieve packets.

-- 
Marilyn Monroe  
Bring Back the Goddess Inc.
_______________________________________________________________________
The views expressed here are mine, my own and mine only.
URL- http://monroe.temple.edu

-----------[000492][next][prev][last][first]----------------------------------------------------
Date:      26 May 1994 09:26:05 GMT
From:      etxmesa@eos.ericsson.se (Michael Salmon)
To:        comp.protocols.tcp-ip
Subject:   Re: TCP & UDP

In article <ctcCqDsx7.2H9@netcom.com>
ctc@netcom.com (ARTHUR HSIAO) writes:
|> I am new to this area, but I have a small question on TCP and UDP.  From what
|> I learned, TCP provides a connection-oriented service, which is like a phone 
|> call, and consequently when the connection is set up, the client doesn't need 
|> to address every message; it simply sends the data over the link.
|> 
|> UDP, however, provides a connectionless service, which is similar to sending 
|> a message by the post office.  Therefore, the client has to address each 
|> message before it's passed to the transport provider.
|> 
|> To my understanding, there is no such distinction between TCP and UDP.  By
|> looking at their headers, I can only feel TCP will guarantee messages would 
|> arrive at the destination in the same order they were sent.  UDP doesn't have
|> such guarantee.

When you make a telephone call very often you speech is packetized and
sent over a network to the destination. The person there doesn't notice
that your voice is comming in packets, with duplicated and skipped
packets in fact. If your call involves a cellular phone then the path
through the network also changes, automagically, without your noticing
it (hopefully anyway). It isn't what happens at the IP or physical
levels that count, it is how the user experiences the connection. In
the packets there isn't much difference between UDP and TCP, it is how
those packets are treated that matters.

Also TCP guarantees that bytes will arrive at the destination
application in the order that the were sent, messages aren't a part of
TCP.

-- 

Michael Salmon

#include	<standard.disclaimer>
#include	<witty.saying>
#include	<fancy.pseudo.graphics>

Ericsson Telecom AB
Stockholm

-----------[000493][next][prev][last][first]----------------------------------------------------
Date:      Thu, 26 May 1994 10:19:58 GMT
From:      olah@cs.utwente.nl (Andras Olah)
To:        comp.protocols.tcp-ip
Subject:   Re: server cannot bind() again, help?

In article <CqDr8s.4w@ucdavis.edu>, heberlei@cs.ucdavis.edu  (Louis Todd Heberlein) writes:
|> I have set up a single-process, concurrent server with TCP
|> sockets, and I am having a problem doing bind() on the second
|> run of the server; I get an address already in use error.

[ detailed description deleted ]
 
|> It appears that a single-process, concurrent server over TCP must get
|> the clients to close() their sockets first before the server does; otherwise,
|> the server will not be able to immediately run again listening to the same
|> port.

You receive `address in use' because the port of the server is in the
TIME-WAIT state.  TCP uses this delayed release of the port to prevent
the reuse of the connection specified by (source IP address, source port,
destination IP address, dest. port) before packets from a previous
incarnation are still in the network.  For this purpose it's enough to
block only one end of the connection (attempts for connecting from the
other end will be rejected).  If you check out the state diagram in p23
of RFC793, you'll see that the side which initiates the close will end up
in the TIME WAIT state.  Therefore, if your server shuts down the
connection, then the server's port will be blocked for 2MSL (~4min).  If
a client initiates the close, then you're not bothered with the TIME WAIT
state, because the port number of the client is not critical, so it just
picks up a different port number.

|> 
|> Can I overcome this?  (I have tried shutdown(), but w/o success).

Have your client close the connection.

|> 
|> Thanks,
|> 
|> Todd Heberlein					heberlei@cs.ucdavis.edu


--  
  Andras Olah                    	olah@cs.utwente.nl

-----------[000494][next][prev][last][first]----------------------------------------------------
Date:      Thu, 26 May 1994 11:41:31 GMT
From:      arcosta@hal02.sarah.br
To:        comp.protocols.tcp-ip
Subject:   Help in MX/SMTP

Hi, 

  I would like to know how to communicate with the SMTP in the MX (VAX/VMS).
  I'm working with software that send and receive mail using port #25.
  I get it to work well with SunOS SMTP (but I don't know how to send the 
subject field :-( .
  But my goal is to do this with the MX SMTP, and I can't even do the
HELO command !!!!
  

  I'd appreciate any help.


  ----------------------------------
  Alexandre Rodrigues Costa
  Sarah Hospital - APS
  Brasilia-DF, Brasil
  E-Mail : arcosta@hal02.sarah.br
           arcosta@icm.sc.usp.br
  ----------------------------------

-----------[000495][next][prev][last][first]----------------------------------------------------
Date:      26 May 1994 20:06:44 -0400
From:      Alan S. Dobkin <ADobkin@EmoryU1.CC.Emory.EdU>
To:        comp.protocols.tcp-ip
Subject:   BootP Server for Macintosh

Does a Macintosh BootP server exist?
If not, what other options do we have to enable BootP with MacTCP?

Any information would be greatly appreciated!

Sincerely,

/------------------------------+---------------------+-------------------\
|  Alan S. Dobkin, Technician  |  2009 Ridgewood Dr  |  E-Mail Address:  |
|  and Student Lab Consultant  |  Uppergate/Griffin  |  ADobkin@EmoryU1  |
|  Emory Univ. Micro. Support  |  Atlanta, GA 30322  |   .CC.Emory.EdU   |
|  Information Tech. Division  |   (404)  727-2766   |  FAX #: 727-0817  |
\------------------------------+---------------------+-------------------/

-----------[000496][next][prev][last][first]----------------------------------------------------
Date:      26 May 1994 13:45:44 GMT
From:      nhs@llnl.gov (Norman H. Samuelson)
To:        comp.protocols.tcp-ip.ibmpc,comp.protocols.tcp-ip,comp.protocols.ppp
Subject:   Re: security of SLIP/PPP

In article <2ruack$1kjg@mtecv2.mty.itesm.mx> mpinones@netmon.mty.itesm.mx (Marco A. Pinones I.) writes:
>Wilfred Mollenvanger (wilfred@zeus.wnc.nedlloyd.nl) wrote:
>How do you protect from your terminal server the slip users? Can you
>assign multiple passwords to one port or accounts?
>Is there any way to accomplish such task? We want to provide our
>developers ppp or slip access to the net, but we want also to protect
>the access.
>Greetings,

The way our SLIP server is setup is fairly simple.  It is a "terminal server".
It handles all the dialups into our systems.  The "ordinary" user logs in with
a username and password, then selects the system to which he wants to connect,
then logs in to that system.  (Think of that like logging in to one machine
then doing TELNET to get to another machine).  The SLIP user logs in the first
time just like the "ordinary" user, then rather than connect to a host, he (or
more likely his software) types "slip <hostname>" (that hostname was assigned
to him along with his IP address), then types his password.  In either case
you wind up giving your password twice.  It is done by a script in either of
the SLIP implementations I have tried on the PC end (Trumpet and Chameleon).

- Norm -

-----------[000497][next][prev][last][first]----------------------------------------------------
Date:      26 May 1994 14:42:41 GMT
From:      hickey@ctron.com (Gerard Hickey)
To:        comp.protocols.tcp-ip
Subject:   Re: Unix computer as a router?

In article <CqD8t1.47D@onyx.indstate.edu> xxmcleis@indsvax1.indstate.edu writes:
>Can this be done? Can anyone give my an example of a computer acting
>as a router, specifically to an Internet provider w/ 56k digital line?
>
>I realize there may be a performance limitation, but we may not have the
>money to get a *real* router like a CISCO. 
>
So you would rather use a $7000-$8000 Unix box as opposed to a $2500 
cisco router. Not to mention that the Unix box would need to be adapted
(i.e. V.35  or RS-449 interface) to the CSU/DSU (about $800).

Unix machines were the first routers, and they worked pretty well. 
But routers provide a lower cost and better performance while making
the box more maintainable.

So, yes, you can use a Unix box can act as a router. In fact, that is
the way it is configured when it comes out of the box.

Good luck...

-- 
					Gerard.
-------------------------------------------------------------------
Gerard Hickey				hickey@ctron.com
Cabletron Systems, Inc.			+1 603 337 3650

-----------[000498][next][prev][last][first]----------------------------------------------------
Date:      26 May 1994 16:04:11 GMT
From:      dswartz@pugsley.osf.org (Dan Swartzendruber)
To:        comp.protocols.tcp-ip
Subject:   Re: Unix computer as a router?

In article <2s2ch1INNeqf@ctron-news.ctron.com>, hickey@ctron.com (Gerard Hickey) writes:
> In article <CqD8t1.47D@onyx.indstate.edu> xxmcleis@indsvax1.indstate.edu writes:
> >Can this be done? Can anyone give my an example of a computer acting
> >as a router, specifically to an Internet provider w/ 56k digital line?
> >
> >I realize there may be a performance limitation, but we may not have the
> >money to get a *real* router like a CISCO. 
> >
> So you would rather use a $7000-$8000 Unix box as opposed to a $2500 
> cisco router. Not to mention that the Unix box would need to be adapted
> (i.e. V.35  or RS-449 interface) to the CSU/DSU (about $800).

You don't procure systems for the government, do you?  $7000-$8000?
You are joking, no?  


-- 

#include <std_disclaimer.h>

Dan S.

-----------[000499][next][prev][last][first]----------------------------------------------------
Date:      Thu, 26 May 1994 16:15:39 GMT
From:      ronf@phx.sectel.mot.com (Ron Feigen)
To:        comp.protocols.tcp-ip
Subject:   Re: server cannot bind() again, help?

Use sockopt SO_REUSEADDR ot SO_LINGER
In article 4w@ucdavis.edu, heberlei@cs.ucdavis.edu  (Louis Todd Heberlein) writes:
>I have set up a single-process, concurrent server with TCP
>sockets, and I am having a problem doing bind() on the second
>run of the server; I get an address already in use error.
>
>I have the following situation:
>    if the server wants to quit (for whatever reason), I call close() on
>    all sockets, including the socket which is doing listen() for the server.
>    This causes all the clients to shutdown (also calling close() on all
>    their sockets).
>
>    However, if I try to run the server again, listening
>    to the same port, I get the bind() error. (If I wait a few minutes,
>    I can run the server again and the port will be available.)
>
>
>On the other hand,
>    if all the clients quit first (closing their sockets), and then
>    the server finishes (closing its sockets), it can call bind()
>    on a run immediately after.
>
>It appears that a single-process, concurrent server over TCP must get
>the clients to close() their sockets first before the server does; otherwise,
>the server will not be able to immediately run again listening to the same
>port.
>
>Can I overcome this?  (I have tried shutdown(), but w/o success).
>
>Thanks,
>
>Todd Heberlein					heberlei@cs.ucdavis.edu





-----------[000500][next][prev][last][first]----------------------------------------------------
Date:      26 May 1994 16:20:06 GMT
From:      kjetil@rigel.spd.eee.strathclyde.ac.uk (Kjetil Rossavik)
To:        comp.protocols.tcp-ip
Subject:   V.24

Hi,
Here's a very basic questions for you gurus:

How do I make my sun workstation speak V.24. Is there already a v.24 interface
on it (I wouldn't recognise it if there was :-) ? Which protocol(s) would I
run between V.24 and IP?

Regards,
Kjetil.

-----------[000501][next][prev][last][first]----------------------------------------------------
Date:      Thu, 26 May 1994 16:41:45 GMT
From:      heberlei@cs.ucdavis.edu  (Louis Todd Heberlein)
To:        comp.protocols.tcp-ip
Subject:   Re: server cannot bind() again, help? (THANKS!)

Many thanks to everyone who replied!  I now understand
the problem and have a couple of solutions.

Thanks again!

Todd					heberlei@cs.ucdavis.edu

-----------[000502][next][prev][last][first]----------------------------------------------------
Date:      Thu, 26 May 94 18:32:27 GMT
From:      brian@ilinx.com (Brian J. Murrell)
To:        comp.protocols.tcp-ip,comp.protocols.misc
Subject:   routing with UNIX as a router

I have a relatively simple routing problem.  I have a UNIX system
which I want to set up as a router.  It has two interface cards,
one is a token ring and the other is an ethernet card.  We also
have a class B network which we've carved up into multiple class
C networks.  Here is a diagram of the router...

                       ----------------
    123.45.67.* token  |              | ethernet
   <-------------------| UNIX system  |---------------->
                 ring  |              | networks 123.45.*.*
                       ----------------

we want any packet which does not match 123.45.67.* but matches 123.45.*.*
to get routed from the token ring network to the ethernet network, without
routing anything else.  i.e. a packet destined for 98.76.54.32 should not
get sent out to the ethernet side of the "UNIX router", but should just
die.

if we call the token ring interface tr0 and the ethernet interfacet enet0,
the ip address of enet0 is 123.45.12.1, and we set up a route like 

Destination      Gateway            Flags  Refs   Use      Interface
123.45           123.45.12.1        U      0      0        enet0

will any packets on the token ring network (123.45.67.*) get sent through
the gateway (UNIX system) as they do match 123.45.*.*, or will the UNIX
host realize that they came from the token ring network and leave them there.
i.e. not pass them on to the ethernet side of the network.

b.

-- 
Brian J. Murrell                                               brian@ilinx.com
InterLinx Support Services, Inc.                              brian@wimsey.com
North Vancouver, B.C.                                             604 983 UNIX
        Platform and Brand Independent UNIX Support - R3.2 - R4 - BSD

-----------[000503][next][prev][last][first]----------------------------------------------------
Date:      26 May 94 19:53:42 GMT
From:      yehavi@vms.huji.ac.il
To:        comp.protocols.tcp-ip
Subject:   Re: server cannot bind() again, help?

> I have set up a single-process, concurrent server with TCP
> sockets, and I am having a problem doing bind() on the second
> run of the server; I get an address already in use error.
>
> I have the following situation:
>     if the server wants to quit (for whatever reason), I call close() on
>     all sockets, including the socket which is doing listen() for the server.
>     This causes all the clients to shutdown (also calling close() on all
>     their sockets).
>
>     However, if I try to run the server again, listening
>     to the same port, I get the bind() error. (If I wait a few minutes,
>     I can run the server again and the port will be available.)
> ...

The solution to it is to set the REUSEADDR flag to the socket (it should be
some IOCTL() call).

-----------[000504][next][prev][last][first]----------------------------------------------------
Date:      Thu, 26 May 1994 20:47:40 GMT
From:      rdavis@efn.org (Reid Davis)
To:        comp.protocols.tcp-ip
Subject:   HELP:  Looking for shareware alternative to MacTCP

Howdy, sorry for the (as if they aren't common) question, but does 
anybody happen to know if there is a shareware alternative to MacTCP, and 
if so, where it can be FTP'd from...  It needs to be bug-free w/ Mosaic, 
Nutius, Fetch, Anarchie, etc....  If anybody happens to know of such a 
thing, any info would be appreciated..  Please reply via e-mail..  Thanx 
in advance...


Reid Davis
rdavis@efn.org      {or, not sure from transitions}
rdavis@efn.efn.org  {please try both..            }


-----------[000505][next][prev][last][first]----------------------------------------------------
Date:      Thu, 26 May 1994 21:24:21 GMT
From:      rad@tyrell.net (Bob Daniel)
To:        comp.protocols.tcp-ip
Subject:   Re: Non-Interactive input to telnet

Tim Cullen (timc@sni.com.au) wrote:
: This question has probably been already addressed in a FAQ.
 
: How can I use telnet with input from a file (on a Unix host) ?
 
: i.e telnet host < cmdlist

I don't think this can be done but if you are trying to automatically
login, you can use 'rlogin -l loginid hostname' to login automatically.
A .rhosts file will need to be setup at the host in the users home
directory in order to login w/out a password.

-----------[000506][next][prev][last][first]----------------------------------------------------
Date:      26 May 94 13:49:21 +0500
From:      ron@alpha.nsula.edu (Ron Wright)
To:        comp.protocols.tcp-ip
Subject:   Need Info on Subnetting Class-C Addresses

Can anyone point me to an documents explaining and or illustrating
the subnetting of class 'c' addresses.  I have read through RFC950
regarding subnetting but would like to find some additional materials.

Thanks,

Ron 

--------------------------------------------------------------------------------
  =====  =====  =====   | Ron Wright                    | Voice: (318) 357-5594
   | |    | |    | |    | User Support Specialist       |
 N|    |S|    |U|    | Northwestern State University | Fax:   (318) 357-5745
   | |    | |    | |    | Computer Center               |
  =====  =====  =====   | Natchitoches, Louisiana       | Ron@alpha.nsula.edu
--------------------------------------------------------------------------------

-----------[000507][next][prev][last][first]----------------------------------------------------
Date:      27 May 94 00:18:07 GMT
From:      jbrowne@void.ncsa.uiuc.edu (Jim Browne)
To:        comp.sys.mac.comm,comp.protocols.tcp-ip
Subject:   Re: Q: NCSA Tel 2.6 - TCP Error netpush -23008

marilyn@monroe.ocis.temple.edu (Goddess of Macgic!) writes:

>First of all many thanks to J. Browne for crafting such a fine tcp based
>telnet protocol session for the mac, use it all the time.

I didn't create it.  I merely maintain(ed) it.

>This was the first time I encountered an/this error.
>TCP Error:  failed to get TCP Status.
>netpush -23008
 
>Could this be it due to the fact that I am running MacHTTP 1.3 and while
>somebody was accessing my homepage, NCSA Tel 2.6 didn't get enough TCP 
>status to send/recieve packets.

The author of MacBlue Telnet thinks that something in my queue code causes
this problem.  To be honest, I haven't had the time to look at it at all
lately.  The error is relatively harmless by itself.  Unfortunately, it
puts up that dialog which tends to kill everything else running.

(Wow, a content free post.)
-- 
Jim Browne          Random net.person for this season         jbrowne@uiuc.edu
     >> Somewhat on leave from NCSA for the summer.  Mail all Telnet <<
     >> correspondence to mactel@ncsa.uiuc.edu                       <<
            "Bus error, passengers dumped.  Not in service."

-----------[000508][next][prev][last][first]----------------------------------------------------
Date:      Fri, 27 May 1994 00:59:34 GMT
From:      wchen@netcom.com (Wen Chen)
To:        comp.protocols.tcp-ip
Subject:   Max Num of Sockets/Ports?

I am involved in the development of a larg network application which may
require persistent connections among a huge number of processes.  Therefore,
it has become a big concern that our machine may run out of sockets/ports.

sockaddr_in allocates 2 bytes(16bits) for sin_port.  What's the maximum of
ports a system may allow its applications to use.

In Unix, each process has a limited sized file descriptor table, so we can only
have that many socket (descriptors) in each process.  Then, since all the file
descriptor tables are allocated in the Kernel space, there should be an upper
limit of the nubmer of sockets a system can support.  

Rebuilding the Kernel will change some of the upper bounds. But, will this 
reconfiguration allow infinit expansion of our socket allowance.

To sum it up:
	1. Any absolute max for ports?
	2. Any absolute max for sockets in a system?
	3. How to avoid these limits if persistent connections are
		necessary in the design?

 



-----------[000509][next][prev][last][first]----------------------------------------------------
Date:      27 May 1994 13:24:06 -0700
From:      lstowell@pyrnova.mis.pyramid.com (Lon Stowell)
To:        comp.protocols.tcp-ip
Subject:   Re: Two NIC in the same ethernet segment, possible?

>>
>>  This is HIGHLY dependent on the implementation of the IP on the
>>  specific box.  For example, on some Lachman based SVR4 systems you
>>  may discover that packets are coming IN on both interfaces but
>>  are only being sent on ONE of the two...and if it fails the box is
>>  essentially dead.  
>
>If you are using a subnetmask on the host only, then outgoing packets
>should use both interfaces depending on the receivers address. This
>way you can do a static distribution of the load.


   Yes, but in this case the network portion of the IP address would
   by definition be different.  And you have just by definition
   created not one but two IP networks, or subnets if you prefer.

   As far as having two IP and physical interfaces on the SAME IP
   network, you need to know whether your implementation conforms to
   the "Weak ES" or "Strong ES" model as noted in RFC 1122....which
   applies to hosts and not routers or more formally Gateways. 

   There are proprietary implementations out there that load balance
   very nicely with 2 or more physical interfaces that even share a
   single IP address, but these will interoperate (to my knowledge
   anyway) only with other stations having similar features and
   support for the creative reinterpretation needed for this.


-----------[000510][next][prev][last][first]----------------------------------------------------
Date:      Fri, 27 May 94 10:00:33 EST
From:      stein@gcomm.com
To:        comp.protocols.tcp-ip
Subject:   Gopher Server on AIX


MB>I am attempting to create a Gopher server on an IBM RS/6000. ...
MB>Is there an RFC on this somewhere? If anyone has idea where to get
MB>this info, please post or e-mail me.

RFC 1463 is a 16-page document on the Gopher protocol.  You can get it
by anonymous FTP at ds.internic.net in the rfc directory.  (There are
many other ways to get RFC's.)  BTW I found this out by searching through
rfc-index.txt, which is also at that site.  No other RFC's include
"Gopher" in the title.

-- Bob Stein

===============================================================================
| ... The Galacticomm Demo System - 305.583.7808 - Home of The Major BBS! ... |
===============================================================================


-----------[000511][next][prev][last][first]----------------------------------------------------
Date:      Fri, 27 May 94 10:00:34 EST
From:      stein@gcomm.com
To:        comp.protocols.tcp-ip
Subject:   Re: server cannot bind() again, help?


Heberlein:
IO>-> I have set up a single-process, concurrent server with TCP
IO>-> sockets, and I am having a problem doing bind() on the second
IO>-> run of the server; I get an address already in use error.

Olah:
IO>You receive `address in use' because the port of the server is in the
IO>TIME-WAIT state.

IO>-> Can I overcome this?  (I have tried shutdown(), but w/o success).

IO>Have your client close the connection.

Actually, I was able to overcome this by enabling the SO_REUSEADDR
socket option.  If you have WRStevens' Unix Network Programming, look on
page 320.  Setting this socket option on the server side before calling
bind() will allow bind() to snatch an address away from the lingering
remains of a TCP session in the TIME-WAIT state.

The drawback seems to me to be a pretty obscure set of circumstances
where a server shuts down, the client doesn't get the final ACK for some
reason, a new server connection comes up, and the client mistakes the
new server for the old one.  I imagine this would be most likely to
happen if a server site were to shut down a server and install a new
version or configuration, especially for poorly or remotely connected
clients.  You need to carefully consider giving up the TIME_WAIT
safeguard.

-- Bob Stein, Galacticomm, Inc.

===============================================================================
| ... The Galacticomm Demo System - 305.583.7808 - Home of The Major BBS! ... |
===============================================================================


-----------[000512][next][prev][last][first]----------------------------------------------------
Date:      27 May 1994 13:29:34 -0400
From:      jaeger@access.digex.net (Jaeger T. Cat)
To:        comp.protocols.tcp-ip
Subject:   Generic TCP/IP "Sniffer" program?

My company is currently investigating a contract that would require us
to be able to monitor data being transferred on a specific IP addr/Port #.

Is there a tool available that would do this?  I would think one would have
to be root to do this (or own the connection), but we can do that.  I was
trying to avoid having to make kernel mods to get this to work.

I tried looking for the FAQ, but it was not around in the usual places,
and rtfm.mit.edu was full this afternoon....


thanks,

jtc
-- 
Jaeger.. U.Va. Echols '88.  Ain't Holdem great?!?!   Gules, on a bend argent.
Five equilateral triangles, bendwise of the first, voided.  Crest: above a
piers helmet, a death's head, three quarter profile, proper.  Mantling: Gules,
double argent.  Motto: the Greek letters pi, alpha, omega, epsilon, alpha.

-----------[000513][next][prev][last][first]----------------------------------------------------
Date:      Fri, 27 May 1994 12:52:37
From:      ly_v@measurex.com (Van Chinh Ly)
To:        comp.protocols.tcp-ip
Subject:   Telnet client source wanted

Does anyone know where I can get a copy of the source code for ANY telnet 
client?

Van Ly



-----------[000514][next][prev][last][first]----------------------------------------------------
Date:      Fri, 27 May 1994 13:24:03 UNDEFINED
From:      treynold@fred.lasalle.edu (Tommi)
To:        comp.sources.wanted,comp.protocols.tcp-ip,comp.dcom.lans.ethernet
Subject:   Re: PD ethernet analysis software?

In article <twillis.770047039@sol> twillis@drao.nrc.ca (Tony Willis) writes:
>From: twillis@drao.nrc.ca (Tony Willis)
>Subject: PD ethernet analysis software?
>Date: 27 May 94 13:57:19 GMT
>Summary: need PD ethernet analysis software
>Keywords: Ethernet software, public domain, wanted
 
>Hi there
 
>Can anyone steer me toward a good public domain package for
>displaying and analyzing the packet traffic along an ethernet.
>I have tried the 'etherman' package from Curtin University in
>Australia, but it does not seem to find all packets between
>all hosts on my network (or else my network really is f**ked up!)

There is a package called ethload which does the trick excellently.  It runs 
on top of NDIS, packet drivers, etc. etc.

Unfortunately (God, I hate to tease ;-)  I haven't a clue where I got it 
from.... however, a good "archie -s ethload" should do the trick, because I'm 
quite sure that is indeed the filename....

Hope this helps...

---
Tommi

treynold@fred.lasalle.edu

The above are my thoughts; if you don't like them, don't read them!

-----------[000515][next][prev][last][first]----------------------------------------------------
Date:      27 May 1994 09:42:55 +0100
From:      pmiles@tdc.dircon.co.uk (Peter Miles)
To:        comp.protocols.tcp-ip,biz.sco.general
Subject:   How to kill TCP connection in LAST_ACK state

It is possible (and if so how?) to kill off, manually or otheriwse, a tcp 
connection which is the LAST_ACK state (as shown by netstat)? 

The client side does not appear to be sending a FIN before closing 
its side of a telnet/rlogin connection. This is leaving the server with a 
number of connections in LAST_ACK state with sometimes 30K each of bytes 
in the Send-Q (again, according netstat).

I'd like to find a way to zap these connections (with out shutting down 
the rest of tcp/ip). (Unfortunately, I can't do anything about the client 
side of the connection).

This is on a system running SCO UNIX and SCO TCP 1.2.1.

		-- Pete
-- 
Pete Miles			pmiles@dircon.co.uk
				...pipex!dircon!pmiles

-----------[000516][next][prev][last][first]----------------------------------------------------
Date:      Fri, 27 May 1994 16:52:10
From:      psilver@smtp.microcom.com (Phil Silver)
To:        comp.protocols.tcp-ip,relcom.comp.tcpip
Subject:   jobs-modem/comm h/w, s/w, firmware, eng & tech supp

Microcom, a leader in development of PC comm. h/w (modems, bridges, routers) 
and PC internetworking s/w (Carbon Copy, LAN Express) has opportunites for 
s/w, h/w, firmware, and tech support engineers at its headquarters in Norwood 
Mass. Opportunities range from entry level to senior and require (or prefer) 
knowledge of some of the following: C, C++, assembler (M68000, Z80, Rockwell 
C19-C39), IPX, SPX, NDIS, TCP/IP, Novell, MS-Windows, DDK, SDK, OWL, T1, etc. 
Contact: Paul Silver - psilver@smtp.microcom.com or fax (617) 551-1100

-----------[000517][next][prev][last][first]----------------------------------------------------
Date:      27 May 1994 12:17:53 GMT
From:      carlson@Xylogics.COM (James Carlson)
To:        comp.dcom.servers,comp.protocols.tcp-ip
Subject:   Re: Routing questions on terminal servers and SLIP/PPP

In article <2s3d9o$d6l@hsc.usc.edu>, hvora@hsc.usc.edu (Heena Vora) writes:
|> Hi all:
|> 
|> I'm trying to get some info on terminal servers and slip/ppp routing.  I'm 
|> looking to get a server that dynamically assigns IP addresses. 
|> 
|> If we got 2 (or more) of such servers, and hooked them to a common
|> modem pool, a user calling in could be connected to one or another 
|> at any time (obviously)
|> 
|> So in my routing tables (on my host), how do I tell it that it needs to 
|> look in 2 gateways (for the 2 servers) to find the SLIP/PPP dialer?  
|> 
|> This may be elementary, but I'm just getting a crash course in tcpip.

Unfortunately, this is anything but elementary.

The easiest way to handle this is to avoid routing tables altogether.
If you set up the remote IP addresses to be on the same subnet as the
dial-in servers, then the server will proxy-ARP for that address.  The
only concern then is the length of time that stale data stays in your
system's ARP caches.  (We're working on this problem -- drop Gary Malkin
(gmalkin@xylogics.com) a line about it.)

The harder way is to have the servers advertise the host or network
route for those remote nodes.  This might not be too great an idea,
since (at least with RIP) good news travels slowly but bad news travels
quickly.  That is, updates of newly reachable routes will propagate
slowly through your net, leading to delays before the connection is
usable.  There are ways to organize things to avoid this problem, but it
takes careful network design.  (We're also working on this; same contact
for more information.)

--
James Carlson <carlson@xylogics.com>            Tel:  +1 617 272 8140
Annex Software Support / Xylogics, Inc.               +1 800 225 3317
53 Third Avenue / Burlington MA  01803-4491     Fax:  +1 617 272 2618

-----------[000518][next][prev][last][first]----------------------------------------------------
Date:      27 May 94 12:58:44 GMT
From:      ae6244@leibniz.math.usma.edu (MARKERT ERICH L. MR.)
To:        comp.protocols.tcp-ip
Subject:   Re: WINQVTNET Users

Al,

I am assuming that you are trying to send the .tmp file to a UNIX
printer, since you posted this to a UNIX group as well. 

With that assumption made what you have to ensure is that the .tmp file
is in encapsulated postscript format.  Then using the lpr function of WinQVT
you can send this to your UNIX postscript printer.  One thing that you have
to have your systems administrator do is add each machine's host name to your
UNIX host's /etc/hosts.lpd file.

Erich


In article <2s28eh$9dj@search01.news.aol.com>, alfredh806@aol.com (AlfredH806) writes:
|> We are working with WNQVTNET version 3.96 and are enjoying the
|> product but are having a
|> problem in the area of printing.
|> 
|> We wish to send everything via transparent print because we are
|> formatting all of the data
|> in the billing system prior to printing using ESC sequences. 
|> However, when we do this, all of the
|> output goes into a ~xxxxxxx.tmp file in the Windows directory and the
|> job never makes it to the
|> printer.
|> 
|> When using print manager the print manager icon pops up on the bottom
|> of the screen but when you open it, there are no jobs in the queue.
|> 
|> We are a little lost at this point and would appreciate any
|> assistance WINQVTNET users may provide.
|> 
|> Sincerely,
|> 
|> 
|> 
|> Alfred S. Harding
|> ComputerLand Medical Consulting

-----------[000519][next][prev][last][first]----------------------------------------------------
Date:      27 May 1994 13:13:19 GMT
From:      edm@harpo.dev.uga.edu (Ed Maioriello)
To:        comp.protocols.tcp-ip
Subject:   Re: BootP Server for Macintosh

In article <2s3dik$lue@emoryu1.cc.emory.edu>,
Alan S. Dobkin  <ADobkin@EmoryU1.CC.Emory.EdU> wrote:
>Does a Macintosh BootP server exist?
>If not, what other options do we have to enable BootP with MacTCP?
>
>Any information would be greatly appreciated!
>
>Sincerely,
>
>/------------------------------+---------------------+-------------------\
>|  Alan S. Dobkin, Technician  |  2009 Ridgewood Dr  |  E-Mail Address:  |


BootP should not be specific to any platform so any bootp server should
do the trick.  I bootp many Macs from Novell Netware servers running the
Hellsoft Bootpd.  The only problem I've found is that while MacTCP will
let you bootp (select the "server" button) it still requires you to
enter a gateway address staticly.

Hope this helps,

Ed Maioriello                                         edm@eris.ucns.uga.edu
University Computing & Networking Services            edm@harpo.dev.uga.edu
University of Georgia  ----------------------------------------------------
Athens, Ga. 30602      | First Rule of Troubleshooting: 
(706)-542-6468         |                     If it don't work - plug it in! 




-----------[000520][next][prev][last][first]----------------------------------------------------
Date:      Fri, 27 May 94 13:32:03 GMT
From:      egn@athen.mch.sni.de (Emil Naepflein)
To:        comp.protocols.tcp-ip
Subject:   Re: Two NIC in the same ethernet segment, possible?

In article <2s2hpm$nii@pyrnova.mis.pyramid.com> lstowell@pyrnova.mis.pyramid.com (Lon Stowell) writes:
>In article <2rrle2$ltt@nuscc.nus.sg> ccetc@nusunix1.nus.sg (SuperFast) writes:
>>
>>Is it possible to put two Ethernet cards in a UNIX box, and assign them
>>two different IP address for the same subnet?
>
>  This is HIGHLY dependent on the implementation of the IP on the
>  specific box.  For example, on some Lachman based SVR4 systems you
>  may discover that packets are coming IN on both interfaces but
>  are only being sent on ONE of the two...and if it fails the box is
>  essentially dead.  

If you are using a subnetmask on the host only, then outgoing packets
should use both interfaces depending on the receivers address. This
way you can do a static distribution of the load.

Emil Naepflein


-----------[000521][next][prev][last][first]----------------------------------------------------
Date:      Fri, 27 May 1994 18:40:15
From:      CAMARA@novell.wd.cubic.com (Raland Camara)
To:        comp.protocols.tcp-ip
Subject:   Where is the FAQ

Can anyone point me to the ftp site where the FAQ is kept?

Thanks in advance,

Raland

-----------[000522][next][prev][last][first]----------------------------------------------------
Date:      27 May 94 13:57:19 GMT
From:      twillis@drao.nrc.ca (Tony Willis)
To:        comp.sources.wanted,comp.protocols.tcp-ip,comp.dcom.lans.ethernet
Subject:   PD ethernet analysis software?

Hi there

Can anyone steer me toward a good public domain package for
displaying and analyzing the packet traffic along an ethernet.
I have tried the 'etherman' package from Curtin University in
Australia, but it does not seem to find all packets between
all hosts on my network (or else my network really is f**ked up!)

Thanks

Tony
_______
Tony Willis

Internet  :   twillis@drao.nrc.ca
Snailnet  :   Dominion Radio Astrophysical Observatory
              P.O. Box 248, Penticton, BC, Canada V2A 6K3
BC Tel net:   (604) 493-2277    Faxnet    :   (604) 493-7767
voicemailnet: (604) 490-4343    Localnet  :   ext 343           

-----------[000523][next][prev][last][first]----------------------------------------------------
Date:      27 May 1994 14:59:24 GMT
From:      longm@firnvx.firn.edu (Michael Long)
To:        comp.protocols.tcp-ip
Subject:   Need Domain Name Server Package?

Does anyone know where I can get a freeware package for DOS, Windows, and Mac
 platform Domain Name Server? 

Thanks Mike...
*****************************************************************************
Michael Long
Sr. Systems Programmer                   \\\\
Florida Information Resource Network/    C-oo
Florida Department of Education            ~
E-mail: longm@firnvx.firn.edu         Phone: (904)487-0911
*****************************************************************************

-----------[000524][next][prev][last][first]----------------------------------------------------
Date:      27 May 1994 22:28:27 -0400
From:      gmichaud@gandalf.ca (Guy Michaud)
To:        comp.protocols.tcp-ip
Subject:   Re: TCP & UDP

ctc@netcom.com (ARTHUR HSIAO) writes:
( ... stuff deleted ...) ...

>To my understanding, there is no such distinction between TCP and UDP.  By
>looking at their headers, I can only feel TCP will guarantee messages would 
>arrive at the destination in the same order they were sent.  UDP doesn't have
>such guarantee.
 
>Please correct me if I am wrong.

UDP doesn't split data into multiple datagrams, and does not keep track of  
datagrams sent for retransmission in case of errors.   Thus, a UDP header 
is shorter than a TCP header.

Since TCP is connection-oriented, as you stated, it contains additional
overhead (6 bit flag field) to negotiate a connection.  Also, the
negotiated-connection continues with the Window field (16 bits) providing flow
control credit allocation. 

This confirms your thoughts on this, but remember that certain applications
will always use TCP (telnet, ftp), while others will use UDP (tftp, snmp).
For a better description than these few lines, see RFC 793 and  768.

----
Guy


-----------[000525][next][prev][last][first]----------------------------------------------------
Date:      27 May 1994 22:36:16 -0400
From:      gmichaud@gandalf.ca (Guy Michaud)
To:        comp.protocols.tcp-ip
Subject:   Re: Need Info on Subnetting Class-C Addresses

ron@alpha.nsula.edu (Ron Wright) writes:

>Can anyone point me to an documents explaining and or illustrating
>the subnetting of class 'c' addresses.  I have read through RFC950
>regarding subnetting but would like to find some additional materials.

I don't have anything I can e-mail to you (or post here for that matter),
but I could fax some stuff to you.  Drop me an e-mail at
	gmichaud@charm.gandalf.ca
give me a fax number and preferred send time, and I'll send it.

----
Guy


-----------[000526][next][prev][last][first]----------------------------------------------------
Date:      27 May 1994 22:42:08 -0400
From:      gmichaud@gandalf.ca (Guy Michaud)
To:        comp.protocols.tcp-ip
Subject:   Re: Generic TCP/IP "Sniffer" program?

jaeger@access.digex.net (Jaeger T. Cat) writes:

>My company is currently investigating a contract that would require us
>to be able to monitor data being transferred on a specific IP addr/Port #.
 
>Is there a tool available that would do this?  I would think one would have
>to be root to do this (or own the connection), but we can do that.  I was
>trying to avoid having to make kernel mods to get this to work.

Other than the real thing (NG Sniffer), if you have a Sun machine, you could
look into the etherfind utility.  It depends on your definition of "monitoring
data".

----
Guy


-----------[000527][next][prev][last][first]----------------------------------------------------
Date:      27 May 1994 16:12:57 GMT
From:      cliu@cs.sunysb.edu (Cheng-mean Liu)
To:        comp.protocols.tcp-ip
Subject:   multi-media

Hi Folks:

    Can anyone tell me where can I get information about sending video, voice using TCP/IP protocol?
    Any help will be appreciated.

cliu


-----------[000528][next][prev][last][first]----------------------------------------------------
Date:      27 May 1994 22:53:05 -0400
From:      gmichaud@gandalf.ca (Guy Michaud)
To:        comp.protocols.tcp-ip
Subject:   Re: Help Define Slip

pd@erredfs1.er.att.com (Paul Davidson) writes:


>	Can someone point me to a good article on the functionality
>of SLIP ?
>	Hardware wise Is It
>	PC -> modem <---> modem -> host <---> network
                                   ^^^^
                                   host must support SLIP

>    	Allowing Software wise:
>	ftp-on-the-pc -> modem <--> modem -> host <--> network

	exactly!

>        ie faking tcp/ip functionality vi modem connectivity
>	to a real tcp/ip network ?

	I wouldn't call it faking .. but Serial Line IP.

>Any othe guidance, or book to get ??

	Best place I know is with TCP/IP shops like FTP Software, Wollongong,
	WRQ, Frontier ... even Microsoft.  Also, look for SLIP or DIALSLIP
	available on the net (saw it once on black.ox.ac.uk).


One more thing -- your modems will need to be configured to disable XON/XOFF
flow control... SLIP or PPP requires a clear pipe between you and the SLIP or
PPP server.

----
Guy


-----------[000529][next][prev][last][first]----------------------------------------------------
Date:      27 May 1994 16:15:17 GMT
From:      cliu@cs.sunysb.edu (Cheng-mean Liu)
To:        comp.protocols.tcp-ip
Subject:   ftp

   Hi Folks:

     Does anyone know where I can get the ftp source code?

  cliu


-----------[000530][next][prev][last][first]----------------------------------------------------
Date:      27 May 1994 16:28:28 GMT
From:      dunnce@bcuxs2.bc.edu (Chad A Dunn)
To:        comp.protocols.tcp-ip.ibmpc,comp.protocols.tcp-ip,comp.protocols.ppp
Subject:   Re: security of SLIP/PPP

	I was wondering If anyone could tell me how to enable/disable
the SLIP option on a Synoptics 3395A Terminal Server for security
reasons.

Thanks


-----------[000531][next][prev][last][first]----------------------------------------------------
Date:      Fri, 27 May 1994 17:26:22 GMT
From:      pd@erredfs1.er.att.com (Paul Davidson)
To:        comp.protocols.tcp-ip
Subject:   Help Define Slip


	Can someone point me to a good article on the functionality
of SLIP ?

	Hardware wise Is It

	PC -> modem <---> modem -> host <---> network

    	Allowing Software wise:

	ftp-on-the-pc -> modem <--> modem -> host <--> network
                    

        ie faking tcp/ip functionality vi modem connectivity
	to a real tcp/ip network ?

Any othe guidance, or book to get ??

Thanx All
attmail!pdavidson
 pd@erredfs1.er.att.com (135.102.11.190)
-- 

The Opinions Expressed Are Soley Mine And Do Not
In Any Way Express The Opinions Of My Employer.


-----------[000532][next][prev][last][first]----------------------------------------------------
Date:      27 May 1994 17:44:26 GMT
From:      kannan@cs.ualberta.ca (Kannan Thiruvengadam)
To:        comp.protocols.tcp-ip
Subject:   TLI Vs SOckets

Hello,

Why is TLI (Transport Layer Interface) considered better than 
sockets ? Or is it ?

- Kannan


-----------[000533][next][prev][last][first]----------------------------------------------------
Date:      Fri, 27 May 1994 18:08:01 GMT
From:      jwp@bernie.sal.wisc.edu (Jeffrey W Percival)
To:        comp.protocols.tcp-ip
Subject:   server fails when client crashes...

I've been following the helpful discussion on detecting client crashes,
but can't quite bring all of it together to solve my problem.  Here's
a synopsis:

	client connects to server
	server starts writing periodic messages to client
	client stops reading for a while
	server's select() indicates client not writeable
	client dies
	server's select() indicates client writeable
	server calls write(sock, data, n) where n > 0
	write() fails, DOES NOT RETURN (I just get my unix prompt back
		in the server window.)

What I thought should happen:  socket appears writeable, write() returns
zero, I look at error code to see EPIPE or something like that.  In fact,
however, write() DOES NOT RETURN!  Why is this so?
-- 
Jeffrey W Percival (jwp@larry.sal.wisc.edu) (608)262-8686

-----------[000534][next][prev][last][first]----------------------------------------------------
Date:      Fri, 27 May 1994 18:28:29 GMT
From:      TDTRUE@pucc.Princeton.EDU (Tom True)
To:        comp.protocols.tcp-ip
Subject:   When do I get a SIGIO on a socket?

From the man pages and the sun manuals it is not clear when I will get
a SIGIO, assuming a socket has been set up correctly.  Does it matter
what calls I make?  For instance, if I do a recv()  for some data, will
I get a SIGIO when more data is available?  Or only if I get EWOULDBLOCK?
Or whenever more data is ready (even if I haven't touched what was
already there)?
 
My working assumption si that I only get a SIGIO after I have read in
whatever was on the input stream or after I have gotten an incomplete
send on an output stream.
 
Tom True
Princeton University
CIT - Advanced Applications
TDTRUE@PUCC.PRINCETON.EDU
(609) 258-6064

-----------[000535][next][prev][last][first]----------------------------------------------------
Date:      27 May 1994 18:59:09 GMT
From:      longm@firnvx.firn.edu (Michael Long)
To:        comp.protocols.tcp-ip
Subject:   Re: Generic TCP/IP "Sniffer" program?

In article <2s5alu$ief@access3.digex.net>, jaeger@access.digex.net (Jaeger T. Cat) says:
>
>My company is currently investigating a contract that would require us
>to be able to monitor data being transferred on a specific IP addr/Port #.
>
>Is there a tool available that would do this?  I would think one would have
>to be root to do this (or own the connection), but we can do that.  I was
>trying to avoid having to make kernel mods to get this to work.
>
>I tried looking for the FAQ, but it was not around in the usual places,
>and rtfm.mit.edu was full this afternoon....
>
>
>thanks,
>
>jtc
>-- 
>Jaeger.. U.Va. Echols '88.  Ain't Holdem great?!?!   Gules, on a bend argent.
>Five equilateral triangles, bendwise of the first, voided.  Crest: above a
>piers helmet, a death's head, three quarter profile, proper.  Mantling: Gules,
>double argent.  Motto: the Greek letters pi, alpha, omega, epsilon, alpha.

Jaeger,

There is a product on the market by Network General which will sniff and decode
any of all the Network Protocols that you can think of.  We use it for TCP/IP,
DECNet, Token Ring, Ethernet, HDLC, SNA/SDLC, AppleTalk, etc... you name it is 
does it.

There number is (415) 473-2000.  This is the Cadillac of all the Sniffers on
the market.


Thanks Mike...
*****************************************************************************
Michael Long
Sr. Systems Programmer                   \\\\
Florida Information Resource Network/    C-oo
Florida Department of Education            ~
E-mail: longm@firnvx.firn.edu         Phone: (904)487-0911
*****************************************************************************

-----------[000536][next][prev][last][first]----------------------------------------------------
Date:      27 May 1994 19:16:01 GMT
From:      switt@triceratops.hac.com (Steve Witt)
To:        comp.protocols.tcp-ip
Subject:   IP Loadsharing

First, let me apologize for those other two posts, operator headspace...

I've been reading RFC1131 describing the OSPF routing protocol and I notice 
that it is capable of dicovering and using multiple, equal-cost routes to a 
destination.  This aroused my interest as I'm not sure how multiple routes
are "used" by IP.  

It is my understanding that a routing policy or protocol, like OSPF, executing
as a daemon (such as gated), initializes and updates the IP routing table that
IP uses to forward IP datagrams.  The routes in the IP routing table, 
indicate the next-hop router to which to forward IP datagrams for specific 
destinations (either hosts or networks).  If the routing protocol is 
capable of discovering multiple routes to a destination host or network, is
IP capable of using this information?  It would seem that the IP routing
table would need to have each route from the set of multiple routes to a 
destination, inserted into the table and then IP would 
need to have some method of using each route.  My understanding of current
IP implementations (which is limited to the 4.3BSD from which SunOS was
developed) is that the IP routing algorithm first tries to find a host-
specific route for the datagram, then tries to find a network route for the
datagram, and then uses a default route (if it exists).  It would seem that
if there were multiple entries in the IP routing table for a destination
network or host that the IP routing algorithm would match on the first one
found and forward the datagram accordingly and would not be able to make use
of multiple routes effectively.  Perhaps I have a naive understanding of the
IP routing algorithm.  

Can IP (or particular implementations of IP) perform load sharing given a
routing policy that is capable of discovering useful multiple routes to a 
destination?

-----------[000537][next][prev][last][first]----------------------------------------------------
Date:      Fri, 27 May 1994 19:31:19 GMT
From:      leeson@harrierx.cdev.com (Stan Leeson)
To:        comp.protocols.tcp-ip
Subject:   Releasing Socket after Client Abort

I have a small distributed application that uses socket connections between
a server and one or more clients.  We experience a problem when a client
aborts without closing the socket that it was using.  The socket remains
unused but allocated, and we eventually run out of sockets.  How can this
server detect that the client has aborted and free the socket??

Any help you can offer would be greatly appreciated.  This must be a well
know technology.


-- 
*~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~*~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~*
|     Stan Leeson                        | Telephone: (612) 921-6811        |
|     Computing Devices International    | Fax:       (612) 921-6869        |
|     stan.d.leeson@cdev.com             |                                  |

-----------[000538][next][prev][last][first]----------------------------------------------------
Date:      Fri, 27 May 1994 19:37:51 GMT
From:      bsmith@rahul.net (Bob Smith)
To:        comp.protocols.tcp-ip
Subject:   UDP broadcast?

I am trying to broadcast a differential GPS info to nodes on
a wireless LAN. Can you recommend a good book on UDP broadcast?

Is there a simple source code example available?

Thanks,
-- 
Bob Smith <bsmith@rahul.net>

-----------[000539][next][prev][last][first]----------------------------------------------------
Date:      Fri, 27 May 1994 20:53:44 GMT
From:      leisner@gnu (Marty Leisner 25733)
To:        comp.protocols.tcp-ip
Subject:   Re: Generic TCP/IP "Sniffer" program?

Jaeger T. Cat (jaeger@access.digex.net) wrote:
: My company is currently investigating a contract that would require us
: to be able to monitor data being transferred on a specific IP addr/Port #.
 
: Is there a tool available that would do this?  I would think one would have
: to be root to do this (or own the connection), but we can do that.  I was
: trying to avoid having to make kernel mods to get this to work.
 
: I tried looking for the FAQ, but it was not around in the usual places,
: and rtfm.mit.edu was full this afternoon....


Will tcpdump do what you want (assuming you have a bsd machine where you
can open up /dev/nit?)

--
marty
leisner@sdsp.mc.xerox.com  
Member of the League for Programming Freedom
I don't make many mistakes, but when I make one, its a beaut
				Fiorello LaGuardia

-----------[000540][next][prev][last][first]----------------------------------------------------
Date:      Fri, 27 May 94 20:48:50 BST
From:      db15@ukc.ac.uk (Damiano Bolla)
To:        comp.protocols.tcp-ip
Subject:   IPP Routing Architecture

This report discusses the IPP routing architecture. What follows is
an abstract of it. This is the second version of the paper.
Can you please give me an opinion about it ?
You can find the Postscript version on host:
eagle.ukc.ac.uk
Using anonymous ftp under the directory:
pub/db15/v3

--------------------------------------------------
Abstract

This paper describes the structure of the IPP routing architecture.
IPP is an evolution of the TCP-IP protocol. The main advantage of IPP
is the variable length addressing scheme. An IPP address can be fifteen
bytes long and is optimised for local area networks. The logical
structure of the address is very similar to IP, with the main
difference, that each byte specifies a subnet, apart from the last byte
that indicates a host.
Routing speed is maximised by the fact that all routing tables are
accessed using a direct lookup method. The size of the routing tables
within a router is fixed and small. The above two points allow the
construction of very cheap and fast routers.
This routing architecture supports broadcast, multicast and real time
data. It uses  different routing priorities for each type of service.
This results in a better management of network links. For normal network
traffic the link usage is maximised by automatically load balancing the
usage of available links. A working router can be found in the part of
this document describing the tests done on this architecture.
IPP aims to keep all the other qualities of IP Eg: the method used to
manage flow control, resequencing, etc.. The only things that changes
are the structure of an address, the routing table and routing functions.

Damiano

-----------[000541][next][prev][last][first]----------------------------------------------------
Date:      28 May 1994 01:33:26 GMT
From:      bernt@scam.Berkeley.EDU (Bernt Habermeier)
To:        comp.protocols.tcp-ip
Subject:   reliable udp extension?


Is there a library that makes udp retransmit, order, and throw away duplicate
packets?

I don't want to use tcp because I don't want to have limited number of clients
that can be connected at any one time (I'm running out of file descriptors).


Bernt.

Bernt@xcf.berkeley.edu

-----------[000542][next][prev][last][first]----------------------------------------------------
Date:      28 May 1994 12:07:42 -0500
From:      ROBERTS@DECUS.CA (Rob Slade, Ed. DECrypt & ComNet, VARUG rep, 604-984-4067)
To:        alt.books.reviews,alt.books.technical,biz.books.technical,comp.dcom.telecom.tech,comp.mail.misc,comp.protocols.iso.x400,comp.protocols.tcp-ip,misc.books.technical,vmsnet.mail.misc
Subject:   "The Internet Message" by Rose

BKINTMSG.RVW  940309
 
Prentice Hall
113 Sylvan Avenue
Englewood Cliffs, NJ   07632
(515) 284-6751
FAX (515) 284-2607
phyllis@prenhall.com 
70621.2737@CompuServe.COM Alan Apt
Beth Mullen-Hespe beth_hespe@prenhall.com
"The Internet Message", Rose, 1993, 0-13-092941-7
mrose@dbc.mtview.ca.us
 
Could there be some connection between a cover design strongly reminiscent of
Douglas Adam's, "Long Dark Tea-Time of the Soul" and a banner stating that this
is the fourth book in Marshall Rose's trilogy?
 
For those wanting to know how to use Internet mail, this is not your book. 
This is a technical work examining the design aspects of electronic mail
systems.  The Internet RFC822 and OSI's (Open System Interconnection) MHS
(Message Handling System), aka X.400, are the two major examples used in the
review.  Those who know Rose's views of OSI will know which comes off better.
 
In spite of the strong (and readily admitted) bias, this is a thorough analysis
of a frequently bypassed field.  For those who need to build or design
messaging systems, this is required reading.
 
copyright Robert M. Slade, 1994   BKINTMSG.RVW  940309

==============                      
Vancouver      ROBERTS@decus.ca    | Ceterum
Institute for  Robert_Slade@sfu.ca | censeo
Research into  rslade@cue.bc.ca    | Datapac
User           p1@CyberStore.ca    | delendam
Security       Canada V7K 2G6      | esse

-----------[000543][next][prev][last][first]----------------------------------------------------
Date:      28 May 1994 04:44:37 GMT
From:      M. Brent Sleeper <bsleeper@carleton.edu>
To:        comp.sys.novell,comp.protocols.tcp-ip
Subject:   [?] SLIP, PowerMac, & Novell/TCP-IP Token Ring

[I recently posted this in comp.sys.mac.comm. Belatedly recognizing my
lack of inclusiveness, ;-), I'm posting a copy of my question here.
Thanks for any help!]

Howdy,

Here's yet another variation on the "how can I use SLIP and my local
network..." FAQ. Unfortunately, I need to ask it.

One of my clients has a Novell NetWare token ring network of about 8
MS-DOS
compatibles, and they're thinking about getting a 28.8K SLIP connection
to the
Internet. Being the Macintosh evangelist that I am, I've talked them into
using
a PowerMacintosh as the "gateway" machine. (Partly for my own ease of
use...I'd
be the one responsible for server applications and the like.)

Hooking up the Mac into the network isn't a problem, and neither is
transmitting TCP/IP over NetWare. My question is this: how can we get the
TCP/IP packets from the SLIP line onto the token ring? Our provider has
reserved stable IP addresses for us, if that makes a difference.

Thanks in advance for any tips,
Brento

--
M. Brent Sleeper -- bsleeper@carleton.edu
  Flightpath Communications -- Flightpath@aol.com

-----------[000544][next][prev][last][first]----------------------------------------------------
Date:      28 May 1994 11:56:45 -0400
From:      gmichaud@gandalf.ca (Guy Michaud)
To:        comp.sys.novell,comp.protocols.tcp-ip,comp.sys.mac.comm
Subject:   Re: [?] SLIP, PowerMac, & Novell/TCP-IP Token Ring

M. Brent Sleeper <bsleeper@carleton.edu> writes:

>One of my clients has a Novell NetWare token ring network of about 8
>MS-DOS
>compatibles, and they're thinking about getting a 28.8K SLIP connection
>to the
>Internet. Being the Macintosh evangelist that I am, I've talked them into
>using
>a PowerMacintosh as the "gateway" machine. (Partly for my own ease of
>use...I'd
>be the one responsible for server applications and the like.)
 
>Hooking up the Mac into the network isn't a problem, and neither is
>transmitting TCP/IP over NetWare. My question is this: how can we get the
>TCP/IP packets from the SLIP line onto the token ring? Our provider has
>reserved stable IP addresses for us, if that makes a difference.


If you have a TCP/IP stack on your PCs, on the Token-Ring, and you are using
Novell to do the routing between the Token-Ring and the PowerMac gateway,
there should be no problem.  Novell's RIP should route all IP packets across.

If you do not have any other means of 'routing' between both nets, or if you
are using an external router, it would seem that your problem might be an
IP addressing problem.  Remember that if you are routing IP, both sides of 
the net must have a different IP net number.  

Provide additional details, ie.:

        PC         Novell          --modem-----------/--------->
           \______/      \        /
            T-R           PowerMac---modem----------/---------->
           /   
        PC

Include IP and addresses you are using and sub-net masks.

----
Guy


-----------[000545][next][prev][last][first]----------------------------------------------------
Date:      28 May 1994 07:47:05 GMT
From:      Steinar.Haug@runit.sintef.no (Steinar Haug)
To:        comp.protocols.tcp-ip
Subject:   Re: Generic TCP/IP "Sniffer" program?

> Will tcpdump do what you want (assuming you have a bsd machine where
> you can open up /dev/nit?)

As far as I know, NIT is Sun specific. If he really has a BSD system
he's much more likely to have BPF.

Steinar Haug, SINTEF RUNIT, University of Trondheim, NORWAY
Email: Steinar.Haug@runit.sintef.no

-----------[000546][next][prev][last][first]----------------------------------------------------
Date:      28 May 94 12:42:02 GMT
From:      sej@psycfrnd.interaccess.com (Stephen Johnson)
To:        comp.protocols.tcp-ip
Subject:   External Gateway Protocols

Hi,
I'm looking for newsgroups, mail lists or books that will help
educate me about EGPs they're whys and wherefors. If anybody could
help please replay to sej@interaccess.com.
Thanks



-----------[000547][next][prev][last][first]----------------------------------------------------
Date:      29 May 1994 01:23:41 -0500
From:      666736@acadvm1.uottawa.ca
To:        comp.protocols.tcp-ip,comp.protocols.tcp-ip.ibmpc
Subject:   SLIP.

I don't know if this is an appropriate newsgroup for this. If it is, please
reply through e-mail only, as opposed to the group.

What exactly is SLIP? I want to get access to the Internet through a local
provider, and I want a SLIP connection, but am not sure exactly what it is.

Will it allow me to have my own IP address on the Internet?

If it does allow me to have my own IP address, could I run applications for
MS-DOS, say, such as Mosaic, or other applications, using a SLIP link? Or do
some things require a network card, as opposed to a modem?

What is the process of using a SLIP connection? Do you usually login to a
UNIX account, and once you've entered your password, does it initiate the
SLIP?

If I had a SLIP link, would I, say, be able to install LINUX on my system, and
run something like Netrek under XWindows? Or Mosaic?

Is there a FAQ for SLIP?

If I have SLIP, are there Internet packages (freeware or shareware) for
MS-Windows I can use? I know Mosaic exists. What about things like FTP, or
Telnet for Windows? Would I simply be able to execute telnet for windows, and
it would recognize my SLIP link?

I think my main concern, or misunderstanding, is the difference between a SLIP
link, and a link such as an Ethernet link. Obviously, Ethernet is direct, SLIP
uses a serial line, but when it comes down to applications, do they work the
same? Or do you NEED a network card for some things?

Thanks.

-----------[000548][next][prev][last][first]----------------------------------------------------
Date:      Sat, 28 May 1994 17:59:02 GMT
From:      gnn@netcom.com (George Neville-Neil)
To:        comp.protocols.tcp-ip
Subject:   Re: Where is the FAQ

CAMARA@novell.wd.cubic.com (Raland Camara) writes:

>Can anyone point me to the ftp site where the FAQ is kept?

Sure...

ftp.netcom.com:~ftp/pub/gnn/tcp-ip.faq

Enjoy,
George

-- 
gnn@netcom.com

Gentlemen, I will not have you fighting in the War Room.
					--- The President in Dr. Stragelove

-----------[000549][next][prev][last][first]----------------------------------------------------
Date:      29 May 1994 02:57:11 -0500
From:      tar@nbc.ksu.ksu.edu (Tim Ramsey)
To:        comp.unix.solaris,comp.sources.wanted,comp.protocols.tcp-ip
Subject:   Wanted: traceroute for Solaris 2.3

I'm looking for source to traceroute that builds and runs under Solaris
2.3.  I have a traceroute package with RCS header

  static char Version[] = "@(#)traceroute.c       e07@nikhef.nl (Eric Wassenaar) 930211";

but this coredumps on my 2.3 box.  Before I spend time digging at the code
to trace down the bug, I'd rather see if the net has done the work for
me.  A check of archie showed no traceroute tar files newer than ~1991.

Please post or email; thanks in advance.

-- 
Tim Ramsey, tar@ksu.ksu.edu                     Finger for PGP pubkey

-----------[000550][next][prev][last][first]----------------------------------------------------
Date:      28 May 1994 20:12:33 GMT
From:      mjr@tis.com (Marcus J Ranum)
To:        comp.protocols.tcp-ip
Subject:   Re: reliable udp extension?

>Is there a library that makes udp retransmit, order, and throw away duplicate
>packets?

	I suggest you take a look at Sun RPC. 

mjr.

-----------[000551][next][prev][last][first]----------------------------------------------------
Date:      29 May 94 02:35:33 GMT
From:      visser@convex.com (Lance Visser)
To:        comp.protocols.tcp-ip,biz.sco.general
Subject:   Re: How to kill TCP connection in LAST_ACK state

In <2s4bqf$lup@tdc.dircon.co.uk> pmiles@tdc.dircon.co.uk (Peter Miles) writes:

+>It is possible (and if so how?) to kill off, manually or otheriwse, a tcp 
+>connection which is the LAST_ACK state (as shown by netstat)? 

	LAST_ACK means both sides of the connection have sent FINs
to each other.  The side in LAST_ACK is waiting for the FIN it sent
to be ack'ed.

	The only way I can think to kill it off is to somehow forge
a TCP packet with the reset bit set and send it to the connection
stuck in FIN_WAIT from another host.  You could build a command
that would take the necessary ip addresses and tcp ports as arguements
to do it if you had raw access to the network.  It could be done, but
it would be a pain to write.




+>The client side does not appear to be sending a FIN before closing 
+>its side of a telnet/rlogin connection. 

	If that were the case, you would be in ESTABLISHED, FIN_WAIT_1
or FIN_WAIT_2, not LAST_ACK.



-----------[000552][next][prev][last][first]----------------------------------------------------
Date:      29 May 1994 11:57:58 -0400
From:      gwjapan@hamlet.umd.edu (Gateway Japan Project)
To:        comp.protocols.tcp-ip
Subject:   ftp within telnet over ppp?

We are planning to set up a Sun SPARCstation with PPP connection.

I would like to know if there is a way to send files back to callers
by ftp or other TCP/IP method within their telnet sessions.  That is,
could a caller connect to our server, run a shell program on the server,
and have files ftp'd back to them automatically, or the like?

I would appreciate any advice anyone in thes group might offer.


* Larry Garfield, Systems Development
* Gateway Japan ("Opening the Barriers to Information on Japan")
* 1424 16th Street, NW
* Suite 602
* Washington, DC 20036
* tel: 202-884-7647
* fax: 202-265-4673
* biz: gwjapan@hamlet.umd.edu
* fun: lg76@umail.umd.edu


-----------[000553][next][prev][last][first]----------------------------------------------------
Date:      29 May 1994 12:14:28 -0400
From:      unixc@unix1.circ.gwu.edu (CCEW UNIX C Course Account)
To:        comp.protocols.tcp-ip
Subject:   ftp within telnet over ppp?

We are planning to set up a Sun SPARCstation with PPP connection.

I would like to know if there is a way to send files back to callers
by ftp or other TCP/IP method within their telnet sessions.  That is,
could a caller connect to our server, run a shell program on the server,
and have files ftp'd back to them automatically, or the like?

I would appreciate any advice anyone in thes group might offer.

[I am resending this from my GWU account just to ensure posting.
If you wish to send a reply to me personally, please use my
UMD accounts. Thanks.]

* Larry Garfield, Systems Development
* Gateway Japan ("Opening the Barriers to Information on Japan")
* 1424 16th Street, NW
* Suite 602
* Washington, DC 20036
* tel: 202-884-7647
* fax: 202-265-4673
* biz: gwjapan@hamlet.umd.edu
* fun: lg76@umail.umd.edu


-----------[000554][next][prev][last][first]----------------------------------------------------
Date:      Sun, 29 May 1994 12:57:35 UNDEFINED
From:      jlewis@inorganic5.chem.ufl.edu (Jon Lewis)
To:        comp.protocols.tcp-ip,comp.protocols.tcp-ip.ibmpc
Subject:   Re: SLIP.

In article <199405290623.BAA29471@news.cs.utexas.edu> 666736@acadvm1.uottawa.ca writes:

>What exactly is SLIP? I want to get access to the Internet through a local
>provider, and I want a SLIP connection, but am not sure exactly what it is.

SLIP is a protocol that uses serial lines (null modems or regular modems) to 
communicate using the internet protocol.

>Will it allow me to have my own IP address on the Internet?

It will require you to have an IP address.  The SLIP server will have to IP 
addresses assigned to the SLIP lines.  When you start SLIP, the server will 
usually tell you what IP address to use.

>If it does allow me to have my own IP address, could I run applications for
>MS-DOS, say, such as Mosaic, or other applications, using a SLIP link? Or do
>some things require a network card, as opposed to a modem?

If you use a dos SLIP "packet driver" you can then run just about any packet 
driver app.  If you use a Windows Winsock TCP/IP stack that supports SLIP, 
like Trumpet Winsock, then you can run nearly any Winsock app.  The SLIP link 
acts just like ethernet as far as programs are concerned except for the 
massive drop in speed.  On the campus ethernet, I can get 80kb/s or more on 
FTP transfers.  On SLIP with a v.32bis modem with v.42bis compression, I 
usually get about 1.6 kb/s on FTP transfers of compressed files.

>What is the process of using a SLIP connection? Do you usually login to a
>UNIX account, and once you've entered your password, does it initiate the
>SLIP?

This varies greatly.  Your school may have a dialup server that doubles as a 
SLIP and or PPP server.  Someone may have setup a LINUX box with slip server 
software.  If you have the resources and can get the ip addresses from your 
net admin, you could setup your own slip server.  I've done this using ka9q 
nos on an old 386sx.  This system is a SLIP, FTP, "Telnet", and Mail 
server...you can ftp to the host part of my email address to see it.  It will 
probably also become a gopher and HTTP server as soon as I have time to set 
those up.

>If I had a SLIP link, would I, say, be able to install LINUX on my system, and
>run something like Netrek under XWindows? Or Mosaic?

I'm only slightly familar with LINUX and not at all with netrek...but if 
netrek communicates via TCP/IP with other systems, you could most likely use 
it over slip unless the lower transfer rate affects the way netrek runs.


>If I have SLIP, are there Internet packages (freeware or shareware) for
>MS-Windows I can use? I know Mosaic exists. What about things like FTP, or
>Telnet for Windows? Would I simply be able to execute telnet for windows, and
>it would recognize my SLIP link?

Once you have something like Trumpet Winsock loaded, winsock apps run without 
caring if you have ethernet or slip.  Winsock apps for just about anything you 
might want can be gotten freely or cheaply.  FTP, telnet, tn3270, POP3 Mail, 
News readers, Mosaic, gopher, finger, IRC, TALK, etc. are all available.

>I think my main concern, or misunderstanding, is the difference between a SLIP
>link, and a link such as an Ethernet link. Obviously, Ethernet is direct, SLIP
>uses a serial line, but when it comes down to applications, do they work the
>same? Or do you NEED a network card for some things?

Some might say you NEED a network card for Mosaic.  Transfering large 
graphics, sounds, or movies can take quite a while at 1.6 kb/s.  But if you 
have patience it will work over SLIP.






|------------------------------------------------------------------|
| jlewis@inorganic5.chem.ufl.edu   If the first address bounces    |
|                                  let me know, and resend to the  |
| jlewis@chem.ufl.edu              second address.                 |
|                                  If the second address bounces   |
|                                  something's seriously wrong.    |
|                                                                  |
|                   Mime attachments are OK                        |
|__________________________________________________________________|



-----------[000555][next][prev][last][first]----------------------------------------------------
Date:      Sun, 29 May 1994 09:10:26 GMT
From:      tung@eespcc (Assistant)
To:        comp.protocols.tcp-ip
Subject:   FPT with history scrollback?

When I use ftp, it is boring to type some command again and again,
are there any ftp client that has the scroll back history like
tcsh?

thanks					tung@eespcb.ncku.edu.tw


-----------[000556][next][prev][last][first]----------------------------------------------------
Date:      Mon, 30 May 1994 02:41:15 UNDEFINED
From:      oldbear@arctos.com (The Arctos Group)
To:        comp.protocols.tcp-ip,comp.protocols.tcp-ip.ibmpc
Subject:   Re: SLIP.

In article <199405290623.BAA29471@news.cs.utexas.edu> 666736@acadvm1.uottawa.ca writes:
>From: 666736@acadvm1.uottawa.ca
>Subject: SLIP.
>Date: 29 May 1994 01:23:41 -0500
 
>I think my main concern, or misunderstanding, is the difference between a SLIP
>link, and a link such as an Ethernet link. Obviously, Ethernet is direct, SLIP
>uses a serial line, but when it comes down to applications, do they work the
>same? Or do you NEED a network card for some things?

The following excerpt may be of some help in understanding what SLIP is and 
what it is not.  As for the "how to connect" and "what functions can 
I run" questions, you should get hold of the current issue of Internet World 
magazine (June 94) which has a couple of good articles on using Windows to 
connect to the net over SLIP or PPP and on connecting the small business LAN 
to the net.  (Internet World, 11 Ferry Lane West, Westport, Connecticut 06880 
USA.  Tel 203 226-6967.  Email: meckler@jvnc.net) 

Simply put, SLIP is just a way of passing IP packets from your machine, 
whatever it may be (unix, dos, windows, mac), to and from the internet over a 
serial line (dial-up, leased or isdn).



        A DISCUSSION OF SLIP AND PPP INTERNET CONNECTIVITY
        --------------------------------------------------
                         Excerpted from
                  TCP/IP Network Administration
                         by  Craig Hunt
             (c) O’Reilly & Associates, Inc. - 1994


TCP/IP Over a Serial Line

     TCP/IP will run over a wide variety of physical media.  The media 
can be Ethernet cables, as in your local Ethernet, or telephone 
circuits, as in a wide area network.
     Almost all data communications take place via serial interfaces.  A 
serial interface is just an interface that send the data as a series of 
bits over a "single wire," as opposed to a parallel interface which send 
the data bits in parallel over "several wires" simultaneously.  This 
description of a serial interface would fit almost any communications 
interface (including Ethernet itself), but the term is usually applied 
to an interface that connects to a telephone circuit via a modem or 
similar device.  Likewise, a telephone circuit is often called a serial 
line.
     In the TCP/IP world, serial lines are used to create wide area 
networks (WAN).  Unfortunately, TCP/IP has not always had a standard 
physical layer protocol for serial lines.  Because of the lack of a 
standard, many designers choose to use a single brand of routers within 
their WAN to ensure successful physical layer communication.  The growth 
of TCP/IP WANs has led to a strong interest in standardizing serial line 
communications to provide vendor independence.
     Other forces that have increased interest in serial line 
communications are the advent of small affordable systems that run 
TCP/IP, and the advent of high-speed, dial-up modems that can provide 
"reasonable" TCP/IP performance.  When the ARPANET was formed, computers 
were expensive and dial-up modems were very slow.  At that time, if you 
could afford a computer, you could afford a leased telephone line.  In 
recent years, however, it has become possible for a user to own a UNIX 
system [or other TCP/IP supporting system] at home.  In this new 
environment, there is an increasing demand for services that allow 
TCP/IP access over dial-up serial lines.
     These two forces, the need for standardized wide area 
communications and the need for dial-up TCP/IP access, have led to the 
creation of two serial line protocols: Serial Line Internet Protocol 
(SLIP) and Point-to-Point Protocol (PPP).

The Serial Protocols

     Serial Line Internet Protocol was created first.  It is a minimal 
protocol that allows isolated hosts to link via TCP/IP over the 
telephone network.  The SLIP protocol defines a simple mechanism for 
"framing" datagrams for transmission across serial lines.  SLIP sends 
the datagram across the serial line as a series of bytes, and it used 
special characters to mark when a series of bytes should be grouped 
together as a datagram.  SLIP defines two special characters for this 
purpose:

     1.   The SLIP END character, a single byte with the decimal
     value 192, is the character that marks the end of a datagram.
     When the receiving SLIP encounters the END character, it
     knows that it has a complete datagram that can be sent up
     to IP.

     2.   The SLIP ESC character, a single byte with the decimal
     value 219, is used to "escape" the SLIP control characters.
     If the sending SLIP encounters a byte value equivalent to
     either a SLIP END character of a SLIP ESC character in the
     datagram it is sending, it converts that character to a
     sequence of two characters.  The two-character sequences are
     ESC 220 for the END character, and ESC 221 for the ESC 
     character itself.  When the receiving SLIP encounters these
     two byte sequences, it converts them back to single-byte
     values.  This procedure prevents the receiving SLIP from 
     incorrectly interpreting a data byte as the end of the datagram.

     SLIP is described in Internet RFC 1055, "A Nonstandard Transmission 
of IP Datagrams Over Serial Lines: SLIP."  As the name of the RFC makes 
clear, SLIP is not an Internet standard.  The RFC does not propose a 
standard; it documents an existing protocol.  The RFC identifies the 
deficiencies in SLIP, which fall into two categories:

     1.   The SLIP protocol does not define any link control
     information that could be used to dynamically control the
     characteristics of a connection.  Therefore, SLIP systems
     must assume certain link characteristics.  Because of this
     limitation, SLIP can only be used when both hosts know
     each other’s address, and only when IP datagrams are being
     transmitted.

     2.   SLIP does not compensate for noisy, low-speed telephone
     lines.  The protocol does not provide error correction or
     data compression.

     For many applications these problems are unimportant.  You may only 
be interested in sending IP datagrams.  You probably know the address of 
both hosts involved in the link, and can, at worst, provide these 
addresses manually.  Additionally, today’s modems provide their own 
compression and error correction.  Given these conditions, SLIP is 
usually considered adequate for linking isolated hosts.  However, in a 
dynamic environment such as a large WAN, SLIP’s problems are generally 
viewed as major weaknesses, and it is usually considered an inadequate 
protocol for linking routers.
     To address SLIP’s weaknesses, Point-to-Point Protocol (PPP) was 
developed as an Internet standard.  Two RFCs that document PPP are: RFC 
1171, "The Point-to-Point Protocol for the Transmission of Multi-
Protocol Datagrams Over Point-to-Point Links"; and, RFC 1172, "The 
Point-to-Point Protocol (PPP) Initial Configuration Options."  As this 
writing, there are several proposed RFCs that add to the PPP standard.
     Point-to-Point Protocol addresses the weaknesses of SLIP with a 
three-layered protocol:

     1.   Data Link Layer Protocol
          The Data Link Layer Protocol used by PPP is a slightly
     modified version of High-level Data Link Control (HDLC).
     PPP modifies HDLC by adding a Protocol field that allows PPP
     to pass traffic for multiple Network Layer protocols.  HDLC
     is an international standard protocol for reliably sending
     data over synchronous lines.  PPP also uses a proposed
     international standard for transmitting HDLC over asynchronous
     lines.  So PPP can guarantee reliable delivery over any type
     of serial line.

     2.   Link Control Protocol
          The Link Control Protocol (LCP) provides control
     information for the serial link.  It is used to establish the
     connection, negotiate configuration parameters, check link
     quality, and close the connection.  During the parameter
     negotiations, LCP can negotiate compression.  LCP was
     developed specifically for PPP.

     3.   Network Control Protocols
          The Network Control Protocols (NCP) are individual
     protocols that provide configuration and control information
     for the Network Layer protocols.  Remember, PPP is designed
     to pass data for a wide variety of network protocols. NCP
     allows PPP to be customized to do just that.  Each network
     protocol (DECnet, IP, OSI, etc.) has its own NCP protocol.
     The NCP protocol defined in RFCs 1171 and 1172 is the
     Internet Control Protocol (IPCP0, which supports Internet
     protocol.

     PPP is a more robust protocol than SLIP, but it is also more 
difficult to implement and not as widely available.  However, PPP’s 
advantages make it the serial line protocol of the future, and the 
likely choice for router vendors seeking a standard mechanism for 
communicating over serial lines.

Choosing a Serial Protocol

     Many network administrators debate over which serial protocol is 
best.  In reality, the correct choice not always the "best protocol" in 
abstract terms; rather it is the "right protocol" for a specific 
situation.  If you have a large network, you will probably use both PPP 
and SLIP.
     PPP is preferred because it is a proposed Internet standard.  
Therefore, it will ensure interoperability between systems from a wide 
variety of vendors.  Additionally, PPP has more features, and is more 
robust, than SLIP.  These features make PPP a good choice as a non-
proprietary protocol for connecting routers over serial lines.  However, 
because SLIP was the first widely available serial protocol for IP, and 
because it is simple to implement, SLIP is available for more kinds of 
hardware than PPP.
     One of the largest applications for IP over serial lines is dial-up 
access.  SLIP is more widely used for this than PPP is, because many of 
the systems that offer dial-up access only support SLIP.  SLIP is 
available for most terminal servers, and in most PC implementations of 
TCP/IP.  SLIP and PPP do not interoperate; they are completely different 
protocols.  So if your terminal servers only have SLIP, the remote hosts 
that connect through these servers must also have SLIP.  Because of its 
installed base, SLIP will continue to be widely used for the foreseeable 
future.
     So which protocol should you use?  The answer is both.  PPP is the 
serial protocol of the future; use it where you can.  For example, when 
you are designing a new serial line service, try to use PPP.  However, 
you should continue to support SLIP.  SLIP is adequate for most 
applications, and is often the only serial protocol available for a 
specific piece of hardware.  Simply put, use PPP where you can and use 
SLIP where you must.

----------

TCP/IP Network Administration
by Craig Hunt     (472pp)
is available from the publisher
O’Reilly & Associates, Inc.
103 A Morris Street
Sebastopol, California 95472-9902
Tel  800 998-9938  or  707 829-0515
Fax  707 829-0104  Email: order@ora.com

Also of interest from the same publisher:
Connecting to the Internet
by Susan Estrada  (170pp)
  



-----------[000557][next][prev][last][first]----------------------------------------------------
Date:      Sun, 29 May 1994 22:35:56 GMT
From:      ian@pipex.net (Ian Phillipps)
To:        comp.protocols.tcp-ip
Subject:   Re: SUMMARY: FTP problem between Europe and USA

In article <2qvddb$jbe@sheckley.cnam.fr>,
Stephane Bortzmeyer <bortz@cnam.cnam.fr> wrote:
>In article <2qqqdc$qec@sheckley.cnam.fr>, bortz@cnam.cnam.fr (Stephane Bortzmeyer) writes:
 
>Now, the question is: can anynone explain why a f*$#@ problem on a 
>line, instead of resulting in loss packets, delays, etc, results in 
>impossibility of transferring *some* files and not others? And why
>not even one byte is transferred on the offending files?

A possibility, and it's just a guess, is that Renater is using IP
class-of-service to send packets by different routes. Some FTP clients
correctly use class-of-service - asking for low latency on the control
conection, and high throughput on the data connection.

If there were two routes (e.g. a high-bandwidth satellite link and a low
bandwidth terrestrial one) and the satellite failed, you could see this
until/unless the routers noticed the outage or were reconfigured.

That's several shots in the dark, though.

Ian
-- 
Ian Phillipps. Tech support manager, Unipalm. News admin, pipex. Internic: IP4
Many system managers claim that holes in an NNTP stream are more valuable than
the data. Van Jacobson, RFC 1144


-----------[000558][next][prev][last][first]----------------------------------------------------
Date:      30 May 1994 04:57:04 GMT
From:      cliu@cs.sunysb.edu (Cheng-mean Liu)
To:        comp.protocols.tcp-ip
Subject:   ftp

 Hi Folks:
    Can anyone show me how to get information about sending
video, voice using TCP/IP (tcp, or udp)?

  Any help will be appreciated.

  Cheng-mean Liu

-----------[000559][next][prev][last][first]----------------------------------------------------
Date:      30 May 1994 05:16:40 GMT
From:      iversen@dsfys1.fi.uib.no (Per Steinar Iversen)
To:        comp.protocols.tcp-ip
Subject:   Re: FPT with history scrollback?


In article <1994May29.091026.10701@dec8.ncku.edu.tw>, tung@eespcc (Assistant) writes:
>When I use ftp, it is boring to type some command again and again,
>are there any ftp client that has the scroll back history like
>tcsh?
>
>thanks					tung@eespcb.ncku.edu.tw

Try ncftp from cse.unl.edu:/pub/mgleason/ncftp
It supports both scroll back and many other goodies. You will need to 
define either readline or getline support to use the scroll back feature.

-psi

-----------[000560][next][prev][last][first]----------------------------------------------------
Date:      30 May 94 06:33:44 GMT
From:      claude@bauv.unibw-muenchen.de (Claude Frantz)
To:        comp.protocols.tcp-ip
Subject:   Re: Generic TCP/IP "Sniffer" program?

longm@firnvx.firn.edu (Michael Long) writes:

>In article <2s5alu$ief@access3.digex.net>, jaeger@access.digex.net (Jaeger T. Cat) says:
>>
>>My company is currently investigating a contract that would require us
>>to be able to monitor data being transferred on a specific IP addr/Port #.
 
>There is a product on the market by Network General which will sniff and decode
>any of all the Network Protocols that you can think of.  We use it for TCP/IP,
>DECNet, Token Ring, Ethernet, HDLC, SNA/SDLC, AppleTalk, etc... you name it is 
>does it.
 
>There number is (415) 473-2000.  This is the Cadillac of all the Sniffers on
>the market.

We have Network General's Sniffer here, with the software version 3.00.
Now, we need an additional protocol suite but our distributor says that
it is not more available, in contrast to the past statement that we can buy any
protocol suite later. Now we should buy the new software upgrade to version
4.4 for DM 8268.50, i.e. aprox. US$ 5168. At my own opinion, this is not
an interesting offer nor a good solution.

--
Claude F. (claude@bauv106.bauv.unibw-muenchen.de)

This message may contain opinions which are not shared by my employer.
The facts can speak for themselves.

-----------[000561][next][prev][last][first]----------------------------------------------------
Date:      30 May 1994 07:33:13 GMT
From:      eric@darling.csl.sni.be (Eric Vyncke)
To:        comp.sources.wanted,comp.protocols.tcp-ip,comp.dcom.lans.ethernet
Subject:   Re: PD ethernet analysis software?

Tommi (treynold@fred.lasalle.edu) wrote:
: In article <twillis.770047039@sol> twillis@drao.nrc.ca (Tony Willis) writes:
: >Can anyone steer me toward a good public domain package for
: >displaying and analyzing the packet traffic along an ethernet.
: >I have tried the 'etherman' package from Curtin University in
: >Australia, but it does not seem to find all packets between
: >all hosts on my network (or else my network really is f**ked up!)
 
: There is a package called ethload which does the trick excellently.  It runs 
: on top of NDIS, packet drivers, etc. etc.

Two 'official sites':
oak.oakland.edu:/pub/msdos/lan/ethld104.zip
ub4b.eunet.be:/pub/ub4b/network/msdos/ethld104.zip
--
        Eric Vyncke, Project Leader, Postmaster and IP Admin  ;-)
Siemens Nixdorf         -       Centre Software de Liege
Rue des Fories, 2               B-4020 Liege, Belgium
email: vyncke@csl.sni.be        fax: +32.41.201.642	tel: +32.41.201.654
X.400: C=be;A=rtt;P=sni;O=siemens nixdorf;OU1=liege;OU2=l1;OU3=d1;OU4=csl;\
        G=Eric;S=Vyncke

-----------[000562][next][prev][last][first]----------------------------------------------------
Date:      Mon, 30 May 1994 10:27:13 GMT
From:      etxruan@argus30.ericsson.se (Rune Andersson)
To:        comp.protocols.tcp-ip
Subject:   FTP client software

Hello out there,

I want all information on how to get an FTP client
SW, free or commercial, that can offer a nice
API library to my application.

Rune Andersson



-----------[000563][next][prev][last][first]----------------------------------------------------
Date:      Mon, 30 May 1994 10:36:57 GMT
From:      cambler@harp.aix.calpoly.edu (Christopher Ambler)
To:        comp.protocols.tcp-ip
Subject:   Gated Help?

I should start this off with a simple plea --- what's the best place to
learn about gated, other than the SCO TCP/IP manual (latest)? I've got the
syntax down just fine, but I'm still having problems.

Here's the setup -

interface sl0, slip line (via ISDN 38.4KBPS) to router with full internet
interface e3E0, Etherlink III, a local ethernet with 5 or 6 machines

the SCO UNIX box is toys.fubarsys.com, 129.65.100.239, talking to the
router across sl0, which is 129.65.100.242. The house ethernet, I've
set up arbitrarilly as 192.42.42, and given toys another IP of 192.42.42.42
on adapter e3E0 (yes, I know... we're going to apply for a class C once
I get this running, and will not use this on the internet until then, I
promise! :-)). 

All machines on the house ethernet (192.42.42.whatever) can talk to each
other just fine, and can talk to toys, of course. toys can talk to any
machine on the house ethernet, and can talk to any internet host as the
default gateway seems to be working fine. The problem is getting any
machine on the house ethernet to talk to any machine on the internet
through toys. Toys does not route packets out. Obviously, it doesn't route
them in, as hosts on the ethernet have never heard of 192.42.42.whatever,
but it SHOULD work the other way, right?

Can anyone tell me what I'm doing wrong, or point me at a reference that
will teach me gated? I conquered named today, and learned BIND/DNS without
a problem (and have 2 boxes here ready to be name servers) but gated is
giving me pains in the naughtybits. 

Can I not use rip to do this? If I use bgp or egp, do I need to get an AS
number first? What number can I assign for now just to test? Am I overlooking
something? Did I fail to make a sacrafice to the ghods of the net? :-) I was
under the impression that the static default gateway would have done it...

Any help is MUCH APPRECIATED!

My gated.conf looks like this:

	traceoptions internal external route rip update ;

	rip yes ;
	hello no ;
	bgp no ;
	egp no ;

	static {
		default gateway 129.65.100.242 pref 100 ;
	} ;

And the output from gated -c looks like this:

	gated[572] version @(#)2.0.1.14.development memory dump on toys.fubarsys.com at Mon May 30 03:06:50 1994

Task and Timers:

	RIP	Port   520	Socket  7	RtProto RIP

		RIP_Update	last: 03:06:50	next: 03:06:50	interval: 00:00:30

	ICMP	Proto   1	Socket  6	<IPHeader>

	KRT	Proto 255	Socket  5	RtProto Kernel

	IF	Socket  4	RtProto Direct

		IF_Check	last: 03:06:50	next: 03:06:50	interval: 00:01:00

	RT

		RT_Age	last: 03:06:50	next: 03:07:20	interval: 00:00:30


	Timers without tasks:

		Time.Mark	last: 03:06:50	next: 03:06:50	interval: 00:10:00

Task to socket mapping:

	socket: 4	task: IF
	socket: 5	task: KRT
	socket: 6	task: ICMP
	socket: 7	task: RIP

RIP:
	Default metric: 16		Default preference: 100


Interface Stats:

		Interfaces: 2 (does not include loopback interface)

	e3E0	192.42.42.42		Index: 1	Preference: 0	Metric: 0
		Up-down transitions: 0		Flags: <Up Broadcast Interface>
		Broadcast Address:   192.42.42.255
		Net Number:    192.42.42.0		Net Mask:    255.255.255.0
		Subnet Number: 192.42.42.0		Subnet Mask: 255.255.255.0

	lo0	127.0.0.1		Index: 2	Preference: 0	Metric: 0
		Up-down transitions: 0		Flags: <Up Loopback Interface NoAge>
		Net Number:    127.0.0.0		Net Mask:    255.0.0.0
		Subnet Number: 127.0.0.1		Subnet Mask: 255.255.255.255

	sl0	129.65.100.239		Index: 3	Preference: 0	Metric: 0
		Up-down transitions: 0		Flags: <Up PointoPoint Interface>
		Destination Address: 129.65.100.242
		Net Number:    129.65.0.0		Net Mask:    255.255.0.0
		Subnet Number: 129.65.100.242		Subnet Mask: 255.255.255.255

Martians:
		224.0.0.0        mask 255.0.0.0      
		223.255.255.0    mask 255.255.255.0  
		192.0.0.0        mask 255.255.255.0  
		191.255.0.0      mask 255.255.255.0  
		128.0.0.0        mask 255.255.0.0    
		127.0.0.0        mask 255.0.0.0      
Gateways referenced by static routes:
		 129.65.100.242      
Redirects: off
	Preference: 20




Routing Tables:
	Install: yes	Generate Default: no
	Revision: 3
	Hashsize: 64		Hashmask: 003f
	Entries:	2 nets	2 hosts

	0.0.0.0        	mask 0.0.0.0        	hash 0	entries 2	state 
		*Static	Preference: 100
			Gateway: 129.65.100.242 	Interface: sl0
			Flags: <GW>	State: <NoAge Changed Int>
			Metric: 0	Revision: 2

		 Kernel	Preference: 255
			Gateway: 129.65.100.242 	Interface: sl0
			Flags: <GW>	State: <Changed Ext>
			Age: 00:00:00	Max Age: 00:03:00
			Metric: 0	Revision: 1


	129.65.100.242 	mask 255.255.255.255	hash 1	entries 2	state Host
		*Direct	Preference:   0
			Gateway: 129.65.100.242 	Interface: sl0
			Flags: <HOST>	State: <Changed P2P Host>
			Age: 00:00:00	Max Age: 00:03:00
			Metric: 0	Revision: 3

		 Kernel	Preference: 255
			Gateway: 129.65.100.242 	Interface: sl0
			Flags: <HOST>	State: <Changed Host>
			Age: 00:00:00	Max Age: 00:03:00
			Metric: 0	Revision: 1


	192.42.42.0    	mask 255.255.255.0  	hash 42	entries 1	state 
		*Direct	Preference:   0
			Gateway: 192.42.42.42   	Interface: e3E0
			Flags: <>	State: <Changed Int>
			Age: 00:00:00	Max Age: 00:03:00
			Metric: 0	Revision: 3


	127.0.0.1      	mask 255.255.255.255	hash 63	entries 1	state 
		*Direct	Preference:   0
			Gateway: 127.0.0.1      	Interface: lo0
			Flags: <HOST>	State: <NoAge Changed NoAdvise Host>
			Metric: 0	Revision: 3


-- 
++Christopher(); // Christopher J. Ambler (chris@toys.fubarsys.com)

The above verbosity is strictly the opinion of the author, his dogs, various
AI, an ISDN Internet connection, and the occasional Ozric Tentacles CD.

-----------[000564][next][prev][last][first]----------------------------------------------------
Date:      30 May 94 12:06:09 MET
From:      per@front.se (Per Lindberg)
To:        comp.protocols.tcp-ip
Subject:   Re: server cannot bind() again, help?

In article <0766AGKOONKPUCKEMD@gcomm.com>, stein@gcomm.com writes:

> Heberlein:
> IO>-> I have set up a single-process, concurrent server with TCP
> IO>-> sockets, and I am having a problem doing bind() on the second
> IO>-> run of the server; I get an address already in use error.
> 
> Olah:
> IO>You receive `address in use' because the port of the server is in the
> IO>TIME-WAIT state.
> 
> IO>-> Can I overcome this?  (I have tried shutdown(), but w/o success).
> 
> IO>Have your client close the connection.
> 
> Actually, I was able to overcome this by enabling the SO_REUSEADDR
> socket option.  If you have WRStevens' Unix Network Programming, look on
> page 320.  Setting this socket option on the server side before calling
> bind() will allow bind() to snatch an address away from the lingering
> remains of a TCP session in the TIME-WAIT state.
> 
> The drawback seems to me to be a pretty obscure set of circumstances
> where a server shuts down, the client doesn't get the final ACK for some
> reason, a new server connection comes up, and the client mistakes the
> new server for the old one.  I imagine this would be most likely to
> happen if a server site were to shut down a server and install a new
> version or configuration, especially for poorly or remotely connected
> clients.  You need to carefully consider giving up the TIME_WAIT
> safeguard.

I beleive a typical case where you need the SO_REUSEADDR option on
your server's listening socket is when you want to be able to quickly
restart the server after it has crashed.

But then:

- What will happen to existing clients after the server has crashed
  and been quickly restarted? I suppose the best thing would be if
  they are notified with appropriate errnos so that they can take
  appropriate action and then try to reconnect, no?

- What if I try to start the same server when it is already running?
  Would the new server just "steal" the socket? Or will confusion and
  madness be the result?

Per Lindberg (per@front.se)         !      __!__
Front Capital Systems, Box 5727,    !  _____(_)_____  Ceci n'est pas une Piper
11487 Stockholm Sweden. 46-86611510 !     !  !  !

-----------[000565][next][prev][last][first]----------------------------------------------------
Date:      Mon, 30 May 1994 13:53:06 GMT
From:      dox@bln.sel.alcatel.de (Guntram Dox)
To:        comp.protocols.tcp-ip
Subject:   variable subnetting AND static routing, no OSPF .. ?


In rfc1219 it is stated, that variable length subnetting requires "that the
intra-domain routing protocol handle multiple different subnet masks." OSPF
(rfc1583) is such an appropriate routing protocol.

However, is it possible to apply variable length subnetting with static
routing (i.e. without using OSPF or any other dynamic routing protocol) ?
Are there any facts, that would make this not possible ?

For example, there would be a router, that would route static between two 
subnets with different subnetmasks , but from the same A,B or C -Class IP
network. No OSPF would be used. RFC1219 would be applied.

BTW, where/what are implementations of the OSPF protocol ?

Any help very appreciated.
Guntram Dox.
Alcatel SEL Berlin

If you prefer PM, then please mail to 
frohmuel@bln.sel.alcatel.de and NOT to this sender since it is still not
reachable from outside.
 

-----------[000566][next][prev][last][first]----------------------------------------------------
Date:      Mon, 30 May 1994 13:58:56 GMT
From:      dox@bln.sel.alcatel.de (Guntram Dox)
To:        comp.protocols.tcp-ip
Subject:   rfc1219 Implementation ??, (On the Assignment of Subnet Numbers)


rfc1219 "specifies a procedure for assigning subnet numbers that eleminates the
need to estimate subnet size", which leads to variable length subnetting.

Are there any implementations of rfc1219 ? , in other words a tool for
assigning subnetmasks, subnetnumbers, and hostnumbers according to rfc1219 ?

Any help very appreciated.
Guntram Dox.
Alcatel SEL Berlin

If you prefer PM, then please email to 
frohmuel@bln.sel.alcatel.de and NOT to this sender since it is still not
reachable from outside.
 

-----------[000567][next][prev][last][first]----------------------------------------------------
Date:      Sun, 29 May 94 20:28:49 +1100
From:      hamish@cloud.apana.org.au (Hamish Moffatt)
To:        comp.dcom.lans,,comp.dcom.lans.misc,,comp.protocols.tcp-ip
Subject:   Bridge terminal server help/software required


I've obtained an old Bridge Communications 10 port terminal server. It
came without the 3.5" drive; I've obtained one from a PC (720k), and
that seems to connect and work okay.

Now I just need the software for it. Does anyone know where I can get
it? There doesn't seem to be an ftp server at bridge.com or
bridge.co.jp.

Any other information that anyone could provide about this beast would
also be appreciated.


Please email your replies - I don't have the time I'd like to read this
group. :-(


hamish


Hamish Moffatt,                    hamish@cloud.apana.org.au
Cloud Nine BBS, Melbourne.            Fax: +61 (0)3 803 6954
+61 (0)3 803 6954 (v.32bis)                3:632/552@fidonet

-----------[000568][next][prev][last][first]----------------------------------------------------
Date:      30 May 94 18:38:15 GMT
From:      rtkao@remus.rutgers.edu (Richard Kao)
To:        comp.protocols.tcp-ip
Subject:   tcp/ip how to verify a connection


Is there a way a server can detect if a client has gone away?  I am
writing a server and I need to know when the client has died do to
unexpected problems like the pc being turned off or a lockup in
windows.  Is there a C function call that will be able to query if the
client is still there or do I have to put special routines in my code
to send polls back and forth.

-----------[000569][next][prev][last][first]----------------------------------------------------
Date:      30 May 94 19:09:34 GMT
From:      sej@flowbee.interaccess.com (Stephen Johnson)
To:        comp.protocols.tcp-ip
Subject:   Re: variable subnetting AND static routing, no OSPF .. ?

dox@bln.sel.alcatel.de (Guntram Dox) writes:


>In rfc1219 it is stated, that variable length subnetting requires "that the
>intra-domain routing protocol handle multiple different subnet masks." OSPF
>(rfc1583) is such an appropriate routing protocol.
 
>However, is it possible to apply variable length subnetting with static
>routing (i.e. without using OSPF or any other dynamic routing protocol) ?
>Are there any facts, that would make this not possible ?

Yes, static routing will work with variable subnetting. However, it
may be possible to create a network subnetting scheme that would be
impossible route through (static or otherwise). Keep it simple and
you shouldn't have any problems (famous last words :) ).

>For example, there would be a router, that would route static between two 
>subnets with different subnetmasks , but from the same A,B or C -Class IP
>network. No OSPF would be used. RFC1219 would be applied.
 
>BTW, where/what are implementations of the OSPF protocol ?

I think there is a version of gated that understands OSPF.

 

-----------[000570][next][prev][last][first]----------------------------------------------------
Date:      30 May 1994 17:15:06 +0200
From:      waversch@info.vub.ac.be (Wout Verschueren)
To:        comp.protocols.tcp-ip
Subject:   help needed on connection refused

Can anybody tell me a reason why a client receives an error 'connection refused'
from a server when using sockets in the Unix domain. The only obvious reason
i can find is that the server is not running. But the server is still there
at the moment of connection. Following the manual it couldn't be that too many
descriptors are used because then, a timeout error would be generated.
I use this stuff on a Hp 9000/800 machine running Hp-Ux 9.0.

Thanks

Wout Verschueren
waversch@info.vub.ac.be
-- 
Wout Verschueren                   |   -(o) (o)-
                                   |       U
                                   |      \_/     The smiling samoerai.
e-mail: waversch@info.vub.ac.be    |      ***

-----------[000571][next][prev][last][first]----------------------------------------------------
Date:      31 May 1994 06:13:40 -0500
From:      hart@chaos.bsu.edu (Jim Hart)
To:        comp.protocols.tcp-ip
Subject:   encrypted IP

Does anybody have an update on the crypto IP efforts?  I've
heard there are several commercial products, including
weakened U.S. DES versions from Semaphore Technologies,
Motorola, Hughes, and UUNET technologies.  Also, some
higher quality foreign versions.  How many of these products
interoperate across vendors?  Are there any public domain
versions available or in the works?  Any emerging standards,
so that if I wrote my own system it could interoperate with
other crypto IP based clients and servers?

thanks,
Jim Hart
hart@chaos.bsu.edu


-----------[000572][next][prev][last][first]----------------------------------------------------
Date:      Tue, 31 May 1994 00:05:37 GMT
From:      mgc@netcom.com (Michael Courtney)
To:        comp.protocols.tcp-ip
Subject:   Re: TCP & UDP

To clarify the difference between UDP & TCP a few points need to be made. Both
protocols concern themselves with process level communication using port 
numbers
to identify the user on both ends. Addressing is resolved at the IP level 
which
is used to carry both UDP and TCP packets. The major difference between the 
two
is TCP guarantees delivery by the virtue of acknowledgments sent back to the 
sender. It also has a flow control feature will allows each side to 
advertise how much data it can receive next. The cost is additional overhead 
to
perform these functions. UDP on the other hand, behaves in much the same way 
as 
the underlying IP protocol which is connectionless in nature, meaning things can come out of order. UDP requires less overhead than TCP. It does provide 
a 
checksum that can be used to provide data integrity but that all folk's. Any 
other features that you would expect with TCP must be provided by the application.
		Hope this helps.
		    mgc
-- 
                                             mgc@netcom.com

-----------[000573][next][prev][last][first]----------------------------------------------------
Date:      Tue, 31 May 1994 10:05:18 -0500
From:      resnick@uiuc.edu (Pete Resnick)
To:        comp.protocols.tcp-ip
Subject:   Re: HELP:  Looking for shareware alternative to MacTCP

In article <CqFGFI.2ox@efn.org>, rdavis@efn.org (Reid Davis) wrote:

>Howdy, sorry for the (as if they aren't common) question, but does 
>anybody happen to know if there is a shareware alternative to MacTCP, and 
>if so, where it can be FTP'd from...  It needs to be bug-free w/ Mosaic, 
>Nutius, Fetch, Anarchie, etc....

If you plan on paying a shareware fee anyway, why not just buy MacTCP? If
you get Adam Engst's Internet Starter Kit book, it comes with MacTCP and a
bunch of other stuff for $29.

I'm not sure how you can imagine that anyone would have written a plug-in
replacement shareware version of MacTCP for any less; writing a TCP/IP
stack is not exactly trivial, you know.

pr
-- 
Pete Resnick    	(...so what is a mojo, and why would one be rising?)
Doctoral Student - Philosophy Department, Gregory Hall, UIUC
System manager - Cognitive Science Group, Beckman Institute, UIUC
Internet: resnick@uiuc.edu

-----------[000574][next][prev][last][first]----------------------------------------------------
Date:      Tue, 31 May 1994 03:56:27 GMT
From:      georgeb@netcom.com (George H. Bosworth)
To:        comp.unix.sys5.r4,comp.protocols.tcp-ip
Subject:   Location of Netdir Lib

I'm trying to compile an application on Dell SVR4 R2.2 that apparently
needs the netdir library, and I've nm'd all around without finding it.
The closest I come is /usr/include/netdir.h, which contains prototype
declarations for the missing functions.

The ld undefined reference list is: 

  Undefined                                   first referenced
   symbol                                     in file
  netdir_hostentbyname             /usr/lib/libsocket.a(nd_gethost.o)
  netdir_free                      /usr/lib/libsocket.a(nd_gethost.o)
  netdir_getbyname                 /usr/lib/libsocket.a(nd_gethost.o)
  getnetconfigent                  /usr/lib/libsocket.a(nd_gethost.o)
  getnetconfig                     /usr/lib/libsocket.a(accept.o)
  freenetconfigent                 /usr/lib/libsocket.a(nd_gethost.o)
  netdir_getbyaddr                 /usr/lib/libsocket.a(nd_gethost.o)
  endnetconfig                     /usr/lib/libsocket.a(socket.o)
  setnetconfig                     /usr/lib/libsocket.a(accept.o)
  netdir_hostentbyaddr             /usr/lib/libsocket.a(nd_gethost.o)

Can anyone point me to the right library file?  Kindly email.

George

George H. Bosworth                           georgeb@netcom.com
35 Granada Court
Portola Valley CA 94028                      415/851-3304

-----------[000575][next][prev][last][first]----------------------------------------------------
Date:      31 May 1994 05:58:54 GMT
From:      mathias@solomon.technet.sg (Mathias Koerber)
To:        comp.protocols.tcp-ip
Subject:   packet drop

I found a peculiar problem today, involving
        a netblazer PN1 (running 2.3)
        two ICL DRS/6000 (one old 740 running 6.5.02, onw newer L300 running 7.2
2)


when i ping the netblazer from the DRS systems, I see a packet loss of
up to 100% (over 1 minute), but mostly around 30%.At the same time I can ping
theblazer from any PC or my two IBM RS/6000s without packet loss.

If I telnet to the blazer from the ICL's, I have problems (long delays, hangs),
but from my PC (Linux) or the IBMs there are no problems.

My network is a Thin-Ethernet.

Does anyone have an idea what could be going on? I sadly have no
network analyzer or so to sniff around myself.

Mathias
--
Mathias Koerber                                       Tel: +65 / 778 00 66 x 29
SW International Systems Pte Ltd                           Fax: +65 / 777 94 01
14 Science Park Drive #04-01 The Maxwell     e-mail: Mathias.Koerber@swi.com.sg
Singapore 0511               http://www.swi.com.sg/public/personal/mathias.html
	Pro is to Con as Progress is to Congress
					- Ambrose Bierce (?)

-----------[000576][next][prev][last][first]----------------------------------------------------
Date:      31 May 1994 06:22:07 GMT
From:      amyrick@clark.net (Witch's Hair)
To:        comp.protocols.tcp-ip
Subject:   I need a FAQ that will explain

Here is what I want to do..... I want to provide a free Dial-up service 
to the internet.  Now here is the questions...

1.  Okay I know I can phone the phone comapnies and they will set up the 
56K or T1 line....  I need to know what hardware is nessary to acomplish 
my task.

Here is what I likes to do.... have the phone Company handle all my 
ringdown lines (I just have one or two physical connections) and if they 
can actually be a modem too.  Have them routed into my system.. Then have 
my system routed via TCP/IP 56K or T1 to the internet.  Now I would like 
to avoid buying 100 modems ans a trillion digiboards to connect to the 
PC.  So if anyone has a bright idea or a good FAQ pleas e-mail me.

--
 ________________________________________________________________________
| Todd Allen Myrick               | Best Friends Computer Company        |
| amyrick@clark.net               | 9204 Santa Rita Rd.                  |
| aka. Witch's Hair on WWIV2@4074 | Voice/FAX (410) 256-0823             |
| "Honor to the Queen of Spades"  | Data: Home By The Sea (410)-256-3792 |
 ------------------------------------------------------------------------

-----------[000577][next][prev][last][first]----------------------------------------------------
Date:      31 May 1994 06:51:35 GMT
From:      patko@uclink.berkeley.edu (Patrick Chung-Pui Ko)
To:        comp.protocols.tcp-ip,comp.protocols.tcp-ip.ibmpc
Subject:   Subnet mask puzzle.Can not find answer in RFCs!!!!

Hello people,
Here is thepuzzle. Is this a valid configuration? CanI  change the
subnet mask within another subnet which has been already subnetted
with a different subnet mask???
 
 
 
original IP address given by NIC is 128.191
subsnetwork 128.191.252, subnet mask 255.255.255.0
====================================
 |
 |
 |subnetwork 128.191.253.0,  subnet  mask  255.255.255.0
 ====================
		 |
		 |
		 + gateway address and mask on each interface
		 + 128.191.253.1, subnet mask 255.255.255.0
		 + 128.191.254.2, subnet mask 255.255.255.128
		 |
		 | subnetwork  128.191.254.0 mask 255.255.255.128
 ===========================================================
		|             |                                           |
hosts    128.191.254.3   128.191.254.4 ...and so until 127                |
									  |
			       		     gateway addresses  and mask  +
					 128.191.254.126, 255.255.255.128 +
					 128.191.254.253, 255.255.255.128 +
									  |
					   sub network 128.191.254.128    |
		       ====================================================
		       host addresses from 128.191.254.128 to 128.191.254.254
 

-----------[000578][next][prev][last][first]----------------------------------------------------
Date:      Tue, 31 May 1994 15:39:39 -0500
From:      wall@cc.swarthmore.edu (Matthew Wall)
To:        comp.networks.noctools.wanted,comp.protocols.tcp-ip
Subject:   SNMP Survey

20 Questions on SNMP for Network Managers

Hello...I am a grad student in Information Systems taking a survey
networking class at Drexel University in Philadelphia. I am doing a
research paper on SNMP for a course and would appreciate any response to
this brief survey, partial or otherwise, from current network managers.
Anything you have to say -- from "SNMP stinks" to your philosophy of life
as it relates to SNMP -- will be most welcome.

The paper discusses the workings of SNMP and various implementations, and
where it seems to be heading in practice. I've done a fair amount of
research derived from the standard printed sources, RFCs, and the SNMP FAQ.
What I'd like to to learn from you is, in a nutshell, how the 'theory' of
SNMP and its implementations in agents and management tools corresponds
with the reality of your present needs, management problems, and standard
procedures.

The results are not going to be tabulated per se, but rather used as
representative opinions to compare and contrast against the written record.
As such, anything you have to say is valuable to me, and the questions (and
a few suggested choices for answers) below are provided only as a reference
frame. Answer as little or as much as you like.

Thanks! 

- Matt

Please send responses via email to 

wall@cc.swarthmore.edu

I apologize if you received this twice -- I apparently misposted two weeks
ago with the "wrong" distribution! (redface)

0. You

( )  may
( )  may not

quote my name and email address in the paper, with the understanding that
this is only for the limited purposes of this single paper and will be read
by only the instructor.

1. Our present use of SNMP can be characterized as:
a) limited use of some agents and management applications
b) use of SNMP to manage bridges, routers, gateways and so forth
c) widespread use of agents on workstation and above-scale computers
d) we use SNMP on everything
e) any other way you'd care to characterize it:

2. We use SNMP-capabale tools
a) on a TCP/IP-only network
b) mixed TCP - other network (which?)
c) on a non-TCP/IP network (how?)

3. We use 
a) an SNMP-based management tool only
b) non-SNMP-based network tools
c) a mixture

(please feel free to elaborate about what you use and don't use)

4. SNMP is _________ managing microcomputer network configurations
a) perfect, I love to use it for...
b) not ready for...
c) often inappropriate for... but I use it anyway...
d) irrelevant for... because we use other tools or do configuration at the
PC and not over the network

5. In what passes for our strategic networking plan, SNMP is:
a) all there is for the time being
b) merely a passing convenience until CMIP or CMIP/CMOT protocols become
standard
c) a small part of a larger network plan
d) irrelevant
e) already passe
f) exciting new technology we hope to implement in the near future

6. The SNMP-based management tools I use are:


7. The SNMP-based agents I use are:


8. The major non-SNMP management tools I use are:


9. This SNMP-based product looks interesting, I'd like to see what it does:


10. What I like about SNMP is:


11. What I don't like about SNMP is:


12. What I like about the current network management tools I have, SNMP or
otherwise, is:


13. What I don't like about these tools I have is:


14. My major concerns in network management are:
(rate, rank, or comment in whatever way you wish)

- fault management of network electronics

- fault management of workstations on the network

- accounting for large pieces of the network

- accounting for individual users of the network

- accounting different types of uses of the network

- security restrictions on access to the network

- security of data on the network

- performance of the network

- configuration and naming of major nodes on the network

- configuration and naming of all nodes on the network, including
microcomputers

15. How do you see SNMP as corresponding to your needs and priorities in
the question above?

16. How big is the network you manage? Does the size and possible growth
affect your use of network tools?

17. What network management issues do you have that aren't addressed by
automated tools?

18. Do you think expert systems will become a big part of your network
management in the next few years? Do you think SNMP-based expert tools will
be part of this?

19. Do you have anything else to add of interest, words of sagacity, etc.?

20. THANKS!!!!!!



-- 
---------------------------------------------------------------
 Matthew Wall * wall@cc.swarthmore.edu * "Whoot! There I am."
---------------------------------------------------------------

-----------[000579][next][prev][last][first]----------------------------------------------------
Date:      31 May 94 07:42:54 GMT
From:      sp@clcs.com.au (Stephen Prince)
To:        comp.unix.programmer,comp.protocols.tcp-ip
Subject:   Need help with XTI over TCP


Hi All,

I have this problem which I'm hoping someone can help me with.
The setup is two SVR4.2 systems using the XTI interface to talk
over TCP. One of the implementations I have no control over.

My server receives a connection via t_listen() and messages
are sent (t_snd) and received (t_rcv) successfully. However,
whilst my server is waiting for another message in t_rcv(),
it receives a disconnect request and hence calls t_rcvdis().

This is the point of confusion, the "reason" code in the t_discon
structure is set to 0x83. Now, the only explanation I can find is
in X.214 (or ISO8072) which says: "Duplicate source reference
detected for the same pair of NSAPs".

Can someone please tell me what this means, and what is causing
it? Also, how can a user control the SRC-REF parameter? Have I
interpreted the "reason" code correctly?

Please respond via email.

Thanks In Advance.

/sp

-----------[000580][next][prev][last][first]----------------------------------------------------
Date:      31 May 94 08:35:45 GMT
From:      visser@convex.com (Lance Visser)
To:        comp.protocols.tcp-ip
Subject:   Re: TCP & UDP

In <mgcCqn49E.Cy0@netcom.com> mgc@netcom.com (Michael Courtney) writes:

+>To clarify the difference between UDP & TCP a few points need to be made. Both
+>protocols concern themselves with process level communication using port 
+>numbers
+>to identify the user on both ends. Addressing is resolved at the IP level 
+>which
+>is used to carry both UDP and TCP packets. 


	If you look at the way it really works, the real "address" for the 
protocols is a port/IP address pair at the TCP or UDP level.  It would
be nice if TCP and IP were the little partitioned blocks everyone draws
in books, but the TCP checksum implementation (a "fake" ip header is
created and folded into the TCP checksum) and demultiplexing issues 
make the port & ip address a meta-address.


+>The major difference between the 
+>two
+>is TCP guarantees delivery by the virtue of acknowledgments sent back to the 
+>sender. It also has a flow control feature will allows each side to 
+>advertise how much data it can receive next. The cost is additional overhead 
+>to
+>perform these functions. 

	The other thing to remember is that TCP is a stream-based 
protocol without record boundaries.  UDP is a message-based protocol.



+>UDP on the other hand, behaves in much the same way 

+>as 
+>the underlying IP protocol which is connectionless in nature, 
+>meaning things can come out of order. 

	The "feature" of UDP being connectionless that is most important
is that you can send and receive from multiple hosts at any time over
the same connection which is not possible over TCP.

	Things can come in out of order, but what tends to happen is
things get lost or dropped.  If you send huge UDP packets that get fragmented
at the IP level, ONE lost IP fragment means the whole UDP packet is
lost.  You could receive 31k of a 32k udp packet (in ip fragments) and if the final
1k gets lost, you have to dump all the data and pretend that none of it arrived.



+>UDP requires less overhead than TCP. 

	Strictly speaking it does, but there are cases where the 
differences are not all that great.  If you send big messages over
UDP, you will hit IP fragmentation which is a substancial overhead.
For ethernet, you hit fragmentation around 1500 byte messages.

	



>It does provide a 
>checksum that can be used to provide data integrity but that all folk's. Any 
>other features that you would expect with TCP must be provided by the application.

	The UDP checksum is optional, which can be a performance win in
certain situations where absolute data integrity isn't necessary.



-----------[000581][next][prev][last][first]----------------------------------------------------
Date:      31 May 1994 12:41:59 GMT
From:      mathias@solomon.technet.sg (Mathias Koerber)
To:        comp.protocols.tcp-ip
Subject:   Wierd netork behaviour

Hi there.
I am at my wit's end.

I have a local network (Thin Ethernet) with
	2 ICL DRS/6000
	2 IBM RS/6000
	1 Linux PC
	~50 DOS/Windows PCS
	1 Netblazer PN1 (V2.3)

Thw whole net in on one line, no subntting etc sone.

since yesterday, I experience a wierd bahaviour.
It seems that the two ICL systems have grave difficulty talking
to the netblazer, while the IBM and the PCs can talk. Then
once in a while I have total outages, and things come
back very slowly. First only the IBMS talk to each other,
then they can alk to one of the ICLs, but none can talk to
the blazer. Later, things change around ...

I see very long return times (4000ms) from one machine to the
blazer, while other have 0. One machine sees ICMP port unreachables
from the other machine while talking to the blazer.

etc, etc, etc.

Sadly, I have no sniffer nor any other analyzer. I have changed a few 
cables and checked  them with a multimeter. that's about all I have.

Has anyone seen problems like these? Any hints?

desperately
Mathias

--
Mathias Koerber                                       Tel: +65 / 778 00 66 x 29
SW International Systems Pte Ltd                           Fax: +65 / 777 94 01
14 Science Park Drive #04-01 The Maxwell     e-mail: Mathias.Koerber@swi.com.sg
Singapore 0511               http://www.swi.com.sg/public/personal/mathias.html
I believe in Spinoza's God who reveals himself in the orderly harmony
of what exists, not in a God who concerns himself with the fates and
actions of human beings.			-- Albert Einstein

-----------[000582][next][prev][last][first]----------------------------------------------------
Date:      31 May 1994 13:33:30 GMT
From:      rene@dhzb.de (Rene Simon)
To:        comp.protocols.tcp-ip
Subject:   long delay TCP

Is there an implementation of a TCP for "long delay" connections like
RFC1323 ?
I need it for SunOS and Solaris.
-- 
Rene Simon				 Telefon:   +49 30 467094-17
Projektgruppe Medizin Informatik         Telefax:   +49 30 4634043
Voltastr. 5, Gebaeude 10, Aufgang 1      E-Mail: rene@DHZB.DE
D 1000 Berlin 65

-----------[000583][next][prev][last][first]----------------------------------------------------
Date:      Tue, 31 May 1994 13:49:50 GMT
From:      francisr@stupid.ucs.indiana.edu (Rob Francis)
To:        comp.protocols.tcp-ip
Subject:   Multiple IP addrs for same hardware interface?


Some told me it's possible to configure an interface say to
have more than one IP address.  As in, when you do an ifconfig.  

I'm wondering what implications this would have on a Unix machine, let
alone if it's do-able or not.

Anyone?

-rob
francisr@indiana.edu

-----------[000584][next][prev][last][first]----------------------------------------------------
Date:      31 May 94 22:57:02 PDT
From:      acblack@vms4.sci.csupomona.edu (Adrian Black)
To:        comp.protocols.tcp-ip
Subject:   Free SLIP or PPP on VAX?

I am wondering if there are any free SLIP programs that run on VMS? My school
does not provide and SLIP service and I was wondering if there was a program I
could run to get a "pseudo" slip or PPP connection? I connect to the VAX using
a 9600bps serial link.

Thanks.

--------
"I fiddled with the thing-a-ma-jig until  \|/  Adrian C. Black
 the doohickey got frazzled and barfed.   -+-  acblack@vms4.sci.csupomona.edu
 NOT!"                                    /|\  CalPoly Uni., Pomona, CA

-----------[000585][next][prev][last][first]----------------------------------------------------
Date:      31 May 1994 16:28:59 GMT
From:      barmar@think.com (Barry Margolin)
To:        comp.protocols.tcp-ip
Subject:   Re: server fails when client crashes...

In article <1994May27.180801.5613@sal.wisc.edu> jwp@bernie.sal.wisc.edu (Jeffrey W Percival) writes:
>	client dies
>	server's select() indicates client writeable
>	server calls write(sock, data, n) where n > 0
>	write() fails, DOES NOT RETURN (I just get my unix prompt back
>		in the server window.)

My guess is that write() is sending a signal (EPIPE, EIO?) instead of
returning an error code, and the signal's default behavior is to kill the
process.  But in this case I would expect the shell in the server window to
give the reason for the process exiting.  However, I think the C shell
doesn't print a message for EPIPE -- it's too common an error (e.g.
"command | head" almost always exits for this reason), so it's probably this.

If your system has a command to trace system calls and signals (e.g.
trace(1) in SunOS and truss(1) in Solaris) you can trace the server process
and see why it's exiting.

-- 
Barry Margolin
System Manager, Thinking Machines Corp.

barmar@think.com          {uunet,harvard}!think!barmar

-----------[000586][next][prev][last][first]----------------------------------------------------
Date:      Tue, 31 May 94 12:06:52 +0400
From:      Oleg Orel <orel@oea.ihep.su>
To:        comp.protocols.tcp-ip
Subject:   Re: FTP client software

Rune Andersson (etxruan@argus30.ericsson.se) wrote:
: Hello out there,
 
: I want all information on how to get an FTP client
: SW, free or commercial, that can offer a nice
: API library to my application.
 
: Rune Andersson


Try use libftp, you can get it from ftp.oea.ihep.su:/libs


-----------[000587][next][prev][last][first]----------------------------------------------------
Date:      31 May 1994 18:48:10 GMT
From:      kwia4000@bronto.zrz.TU-Berlin.DE (Manfred Kwiatkowski)
To:        comp.protocols.tcp-ip,comp.protocols.tcp-ip.ibmpc
Subject:   Re: Subnet mask puzzle.Can not find answer in RFCs!!!!

In article <2sempn$b9r@agate.berkeley.edu>, patko@uclink.berkeley.edu (Patrick Chung-Pui Ko) writes:
> Hello people,
> Here is thepuzzle. Is this a valid configuration? CanI  change the
> subnet mask within another subnet which has been already subnetted
> with a different subnet mask???

If you are able to support it, it _is_ a valid configuration. :-)

Finding a router that supports a non unique subnetmask on its 
interfaces is not a major problem nowadays. The trouble begins
if you want to run a routing protocol. You will have a hard time
finding a RIP-1 implementation that works reasonably in this 
setup. OSPF on the other hand is not (yet?) an out-of-the-box 
protocol suitable for endsystems.

Usually these kind of scenarios pop up if one wants to integrate
PC-LANs (e.g. Netware) into an existing IP environment. Rather than
to make such a topology work I would recommend you 

a) try to get second subnet with a standard mask or if this is
   not an option

b) forget about routing and use a bridge to separate the traffic.

Always consider the task to maintain this sub-subnetwork will not
become easier should this stub ever become multiply connected.
>  
> original IP address given by NIC is 128.191
> subsnetwork 128.191.252, subnet mask 255.255.255.0
> ====================================
>  |
>  |
>  |subnetwork 128.191.253.0,  subnet  mask  255.255.255.0
 ====================
> 		 |
> 		 |
> 		 + gateway address and mask on each interface
> 		 + 128.191.253.1, subnet mask 255.255.255.0
> 		 + 128.191.254.2, subnet mask 255.255.255.128
> 		 |
> 		 | subnetwork  128.191.254.0 mask 255.255.255.128
 ===========================================================
> 		|             |                                           |
> hosts    128.191.254.3   128.191.254.4 ...and so until 127                |
> 									  |
> 			       		     gateway addresses  and mask  +
> 					 128.191.254.126, 255.255.255.128 +
> 					 128.191.254.253, 255.255.255.128 +
> 									  |
> 					   sub network 128.191.254.128    |
> 		       ====================================================
> 		       host addresses from 128.191.254.128 to 128.191.254.254
>  
 
-- 
Manfred Kwiatkowski         kwiatkowski@zrz.tu-berlin.de

-----------[000588][next][prev][last][first]----------------------------------------------------
Date:      Tue, 31 May 1994 18:51:23 GMT
From:      mgc@netcom.com (Michael Courtney)
To:        comp.protocols.tcp-ip
Subject:   Re: Multiple IP addrs for same hardware interface?

Rob Francis (francisr@stupid.ucs.indiana.edu) wrote:

: Some told me it's possible to configure an interface say to
: have more than one IP address.  As in, when you do an ifconfig. 

In general, the anwer to this is no. Each interface must have a unique IP 
addres relative to the set of hosts you are communicating with within your
internet. The exception to this rule maybe a proxy ARP situation which will
allow a host to respond to a special address which is not the same. But to be
honest I don't have any live experience to share on that one.

: I'm wondering what implications this would have on a Unix machine, let
: alone if it's do-able or not.

Duplicate IP addresses can be a real pain and is something you definitely 
want to avoid, because the problem is not all that obvious.


: Anyone?
 
: -rob
: francisr@indiana.edu
-- 
                                             mgc@netcom.com

-----------[000589][next][prev][last][first]----------------------------------------------------
Date:      31 May 1994 19:25:10 GMT
From:      mgb722c@rs730.gsfc.nasa.gov
To:        comp.protocols.tcp-ip
Subject:   Re: Wierd netork behaviour

In <2sfban$4ih@raffles.technet.sg>, mathias@solomon.technet.sg (Mathias Koerber) writes:
>Hi there.
>I am at my wit's end.
>
>I have a local network (Thin Ethernet) with
>	2 ICL DRS/6000
>	2 IBM RS/6000
>	1 Linux PC
>	~50 DOS/Windows PCS
>	1 Netblazer PN1 (V2.3)
>
>Thw whole net in on one line, no subntting etc sone.
>
>since yesterday, I experience a wierd bahaviour.
>It seems that the two ICL systems have grave difficulty talking
>to the netblazer, while the IBM and the PCs can talk. Then
>once in a while I have total outages, and things come
>back very slowly. First only the IBMS talk to each other,
>then they can alk to one of the ICLs, but none can talk to
>the blazer. Later, things change around ...
>
>I see very long return times (4000ms) from one machine to the
>blazer, while other have 0. One machine sees ICMP port unreachables
>from the other machine while talking to the blazer.
>
>etc, etc, etc.
>
>Sadly, I have no sniffer nor any other analyzer. I have changed a few 
>cables and checked  them with a multimeter. that's about all I have.
>
>Has anyone seen problems like these? Any hints?
>
>desperately
>Mathias
If there is no problem with the cable, I would check to make sure that all of 
the TCP/IP parameters are set correctly. If everything is on the same subnet,
each node should have the same broadcast address, subnet mask, etc.

If that turns out OK, then I would add nodes to the network one at a time and see
when the problems start to arise. That should give you a pretty good clue as to
which machine is causing the problem.

Good luck!

/\\yron
*************************************************
*  Myron Bradshaw - Facility Manager                            *
*  Curtis Management Company, Inc.                               *
*  NASA/Goddard Space Flight Center                             *
*  Greenbelt, MD 20771                                                 *
*  E-Mail: mgb@genie.gsfc.nasa.gov, or                            *
*            mgb722c@rs730.gsfc.nasa.gov                           *
*************************************************
*Jesus not only saves, He also frequently makes Backups.*
*************************************************
* Stupid TV! Be more funny!!!                                       *
*                       - Homer Simpson                                *
*************************************************


-----------[000590][next][prev][last][first]----------------------------------------------------
Date:      Tue, 31 May 1994 19:39:49 GMT
From:      jeff@astph (Jeff Martin)
To:        comp.protocols.tcp-ip
Subject:   accept() core dump!


We are running Interactive UNIX, System V, Release 3.2.
We wish to bullet proof our TCP/IP server process from
any and every possible crash.  Currently we are baffled with
a strange one.  If we request an excessive number of client
socket connections, about 60, the accept call will not return
an error, but rather will dump a core.  The stack is printed
below:

        malloc(0x408)   [malloc.c]
        _findbuf(0x4051b0)   [flsbuf.c]
        _filbuf(0x4051b0)   [filbuf.c]
        fgets(0x40462c,0x78,0x4051b0)   [fgets.c]
        rddev(0x4051b0,2,1,0)   [socket.c]
        socket(2,1,0)   [socket.c]
        accept(6,0x7ffffc3c,0x7ffffc38)   [accept.c]
        main(argc=1,argv=0x7ffffc6c)   [db_root.c:82]

It seem that the accept() call is not too accepting.  Has anyone
else noted this behavior and is there something we can do about it.
That is we would preferably desire an error return.

Thanks, Jeff
-- 
Jeff Martin, dbms programmer,		Philadelphia Phillies
INET:	astph!jeff@cse.psu.edu		Voice:	(814)234-8592x32
UUCP:	psuvax1!astph!jeff		FAX:	(814)234-1269
SLOW:	141 West Beaver, Suite A, State College, PA 16801

-----------[000591][next][prev][last][first]----------------------------------------------------
Date:      31 May 1994 20:12:25 GMT
From:      jallen@synopsys.com (Jeff Allen)
To:        comp.protocols.appletalk,comp.protocols.tcp-ip
Subject:   MacIP problem -- no Zones

I am having trouble getting a PowerBook on LocalTalk to talk MacIP
through a FastPath onto our network. Here are the facts I know
from troubleshooting it, which make me really doubt my sanity:

1. Other Macs on the LocalTalk network work fine using the "Server"
   radio button in the MacTCP window.

2. The little "Zones" popup does not show up under the LocalTalk
   icon in the first window after starting MacTCP. All of the other
   Macs on the LocalTalk segment that are working right have the
   zones popup.

3. Appletalk is enabled and working fine on this Mac, since:

4. We reinstalled MacTCP 2.0.4 from the network with no difficulty,
   checked that there was still no Zones popus, then nuked MacTCP DNR
   and MacTCP Prep, and checked again.

Item 1 implies to me that MacIP is working ok on our network and that
the FastPath (for once!) is not to blame. Item 2 is the big problem --
without that zones popup, I don't think it'll work right.

I am at my wit's end on this. Like all my dealings with MacTCP, I am left
with the destinct impression that Apple finally suceeded in writing
self-aware software when writing MacTCP.

Please respond to me. I'll summarize, if necessary.

-- 
--
Jeff Allen	  			    Phone: 415-694-1793
NCS, Synopsys, Inc. Mountain View, CA.	    Email: <jallen@synopsys.com>


-----------[000592][next][prev][last][first]----------------------------------------------------
Date:      31 May 1994 21:03:02 GMT
From:      leonard@telcom.arizona.edu (Aaron Leonard)
To:        comp.protocols.tcp-ip
Subject:   Re: Multiple IP addrs for same hardware interface?


In article <mgcCqoKDo.4Gz@netcom.com>, mgc@netcom.com (Michael Courtney) writes:
|Rob Francis (francisr@stupid.ucs.indiana.edu) wrote:
|
|: Some told me it's possible to configure an interface say to
|: have more than one IP address.  As in, when you do an ifconfig. 
|
|In general, the anwer to this is no. Each interface must have a unique IP 
|addres relative to the set of hosts you are communicating with within your
|internet. The exception to this rule maybe a proxy ARP situation which will
|allow a host to respond to a special address which is not the same. But to be
|honest I don't have any live experience to share on that one.
|
|: I'm wondering what implications this would have on a Unix machine, let
|: alone if it's do-able or not.
|
|Duplicate IP addresses can be a real pain and is something you definitely 
|want to avoid, because the problem is not all that obvious.

I can't really go along with this assessment.  While I believe that
having multiple addresses per interface is a nonstandard thing to 
do on a U*x host, it's an everyday occurrence on cisco routers, and
it's also done (tho unsupported) on VMS+MultiNet systems.

There are a couple of good reasons to have multiple addresses on an
interface:

- to facilitate the use of multiple subnets or networks on a given
  physical cable.  For example, let's say you have 2 class C networks
  which you want to have share a bridged Ethernet.  You can set a 
  cisco (and probably other vendors') router port attached to that
  Ethernet to have an address from each network, so that it can 
  thereby route packets between the networks, so that the stations
  on different nets can communicate.

- to provide a smooth transition in the demise of a well-known host.
  For example, let's say that I have a nameserver with address
  1.2.3.4.  But this box is scheduled for the scrapheap, 
  and will be replaced with 1.2.3.100.  I can configure the
  1.2.3.4 address onto the 1.2.3.100 interface, and so can allow
  for an arbitrarily long period for that address' phaseout, 
  without being required to prop up the old hardware any longer
  than I want to.

- to multiplex services without requiring client specification of 
  alternate TCP ports.  (This is what the Xylogics Annex can do in
  its support of outbound rotaries.)  For example, I may have one 
  system which I want to offer two TELNET services: one which 
  provides a standard login shell, and the other which provides an
  automatic passthru to an outbound X.29 terminal session.  So I
  tell my OS that TELNET connects to the one IP address get hooked
  into the stock login shell, and that connects to the other IP
  address invoke the X.29 client program.  (Alternatively, of course,
  I could provide the latter service by using an alternate TCP port,
  but that's not as nice to my valued users.)

IMO, allowing for greater virtuality and flexibility in using IP
addresses is a Good Thing.  For example, for a multihomed host, it
would be a great thing if it had a virtual "internal" IP address,
which would serve as the endpoint for all of its ordinary IP 
traffic, and which would have "internal" routes to the outside
world thru its various interfaces.  In this manner, TCP connections
on multihomed hosts would be able to ride out interface/network
failures.

Aaron

Aaron Leonard (AL104), <Leonard@Arizona.EDU>
University of Arizona Network Operations, Tucson AZ 85721
  \ Don't lock yourself into open systems. /

-----------[000593][next][prev][last][first]----------------------------------------------------
Date:      31 May 1994 22:49:45 GMT
From:      gadget@engin.umich.edu (gadget)
To:        comp.sys.mac.comm,comp.protocols.tcp-ip
Subject:   Re: Q: NCSA Tel 2.6 - TCP Error netpush -23008

>I didn't create it.  I merely maintain(ed) it.
>
>>This was the first time I encountered an/this error.
>>TCP Error:  failed to get TCP Status.
>>netpush -23008

Think reminds me about another NSCA Telnet bug I run into sometimes.

I've gotten it stuck in a state where it constantly lights up the 
modem send light.

Set up:  NCSA telnet 2.6, macPPP, SupraFax 14.4 v.32

Diagnosis: I can turn the modem on and off, and the send light will
return to solid red as soon as the modem says OK.  It hoses the CPU
and doesn't stop when I quit telnet, I have to reboot.  Usually I can't
even wait long enough to watch it reboot and I throw a hard power restart.
Reminds me of times I turned on br PrimeTime and it single stepped the OS.
VERY SLOW


Anyone else seen this error? its pretty rare, but very annoying.

Tim


-----------[000594][next][prev][last][first]----------------------------------------------------
Date:      Tue, 31 May 1994 23:06:44 GMT
From:      rstevens@noao.edu (W. Richard Stevens)
To:        comp.protocols.tcp-ip
Subject:   Re: server cannot bind() again, help?

>- What if I try to start the same server when it is already running?
>  Would the new server just "steal" the socket? Or will confusion and
>  madness be the result?

SO_REUSEADDR is indeed confusing.  When you specify it before calling
bind(), all it lets you do is bind a local port that is already part
of a more complete association.  The most common case is when you kill
a process that has a TCP connection using that local port, and want to
restart the server before that connection goes through the TIME_WAIT
state.  If I have a server with the socket pair

	{ 128.1.2.3, 21, 192.3.4.5, 1500 }

and kill that process, putting that socket pair into the TIME_WAIT state,
I can specify SO_REUSEADDR and bind the socket { *, 21 }.  That server
should not be able to accept a connection from a client at { 192.3.4.5,
1500 }, since that would duplicate the socket pair that's in the TIME_WAIT
state (assuming 2MSL seconds haven't passed when the client connects).
Unfortunately Berkeley-derived systems do allow this, as long as the ISN
for the new connection is greater than the final sequence number for the
previous incarnation of the connection.

It does *not* let you start multiple occurences of the same TCP server.
Try it and you'll see.  The second instantiation of the server still gets
"address already in use", regardless whether you specify SO_REUSEADDR or
not.  If you're really interested in this topic, I suggest you look at
pp. 242-245 of my recent book "TCP/IP Illustrated" (Addison-Wesley, 1994)
for some additional examples.

Now UDP on a system that supports multicasting (e.g., Solaris 2.x) does
let you start multiple occurences of a server, all reading from the same
local port.  But that's another story ...

	Rich Stevens

-----------[000595][next][prev][last][first]----------------------------------------------------
Date:      Tue, 31 May 1994 23:34:38 GMT
From:      choi@cso.geg.mot.com (Ben Choi)
To:        comp.protocols.tcp-ip
Subject:   How do you determine SO_KEEPALIVE parameters?

Hi,

I've heard that when Unix socket connections are made, there is an option 
that can be used to determine if that connection ever goes down.  
The call is to setsockopt() and the option is SO_KEEPALIVE.  The way 
I heard that this works is
that when this option is set, idle time is measured and if a socket
connection is idle for X amount of time, a TCP message is sent on
the socket and a response is expected back.  If a response is not
received in Y amount of time, the TCP message is sent again.
If no response is received in Z iterations, the socket is closed.

Now here is my question...On the Unix system, would you have any idea
where X, Y, and Z are defined?  

I've looked in W. Richard Steven's Unix Network Programming book but it
seems to leave this detail out.

Any pointers would be appreciated.

Thanks,
Ben




---

 ------------------------------------------------------------------------------
 Ben Choi                                    Internet   : choi@cso.geg.mot.com
 Motorola GSTG Communications Division       Phone      : (602) 441-8268
 8201 E. McDowell Road M/D H8101             Voice Mail : (602) 441-3510
 Scottsdale, AZ 85252                        Fax        : (602) 441-3868
 "Never underpay your software engineers"  - Jurassic Park Moral
 -----------------------------------------------------------------------------



-----------[000596][next][prev][last][first]----------------------------------------------------
Date:      Tue, 31 May 1994 23:56:00 GMT
From:      francisr@stupid.ucs.indiana.edu (Rob Francis)
To:        comp.protocols.tcp-ip
Subject:   Re: Multiple IP addrs for same hardware interface?

In article <2sg8m6$mnn@news.ccit.arizona.edu>,
Aaron Leonard <Leonard@Arizona.EDU> wrote:
>
>While I believe that having multiple addresses per interface 
>is a nonstandard thing to do on a U*x host, it's an everyday 
 occurrence on cisco routers, and it's also done (tho unsupported) 
>on VMS+MultiNet systems.

Agreed.  Indeed cisco routers can do this we do it here.  I'm
wondering about the implications of such and attempt on a network savy
OS such as Unix.  There are several reasons you'd want a router to
have a virtual interface, not as good of reason for a Un*x machinet to
do so, but I'm just looking into the possibilities.



>- to provide a smooth transition in the demise of a well-known host.
>  For example, let's say that I have a nameserver with address
>  1.2.3.4.  But this box is scheduled for the scrapheap, 
>  and will be replaced with 1.2.3.100.  I can configure the
>  1.2.3.4 address onto the 1.2.3.100 interface, and so can allow
>  for an arbitrarily long period for that address' phaseout, 
>  without being required to prop up the old hardware any longer
>  than I want to.


This is basically what I want to do.  Not for nameservice, but I want
to gracefully eliminate a machine but its IP addr is fairly well
distributed.  I actually got around the problem and now I'm just sort
of curious about the whole thing.

-rob
---
Rob Francis, Unix Workstation Support Group, Univ. Computing Services
Indiana University, francisr@indiana.edu (812)855-0327 




END OF DOCUMENT