home *** CD-ROM | disk | FTP | other *** search
- Newsgroups: comp.protocols.tcp-ip
- Path: sparky!uunet!usc!sdd.hp.com!mips!pacbell.com!tandem!dsg.tandem.com!scott
- From: scott@dsg.tandem.com (Scott Hazen Mueller)
- Subject: Confusing SLIP problem
- Message-ID: <1992Jul31.052242.19984@dsg.tandem.com>
- Summary: ping works OK, telnet doesn't
- Keywords: SLIP, ping, telnet, modems, T1600, T3000
- Organization: Tandem Computers Inc., Cupertino CA
- Date: Fri, 31 Jul 1992 05:22:42 GMT
- Lines: 48
-
- I'm having a rather confusing SLIP-related problem, and I'm hoping that
- I've just missed something obvious... The problem is that when I bring
- up my SLIP link, I can do ICMP stuff (e.g. ping, or traceroute from a
- nearby Sun) just fine, but TCP-based apps (telnet, ftp) just say that they've
- connected to the remote, and then hang.
-
- At one end I have Zorch, a 4.3BSD-based system with a fairly antique kernel
- SLIP driver. At the other end is a Xylogics Annex III. I'm using a T1600
- modem at the Zorch end, and a T3000 at the Annex. The T3000/Annex part is
- probably working, as a colleague of mine can SLIP into it with no problems,
- so I assume that the problem is in the Zorch end.
-
- Here's where things start getting confusing:
-
- 1) The SLIP driver on Zorch *seems* OK, though I suppose I could be having
- a header-compression problem, since I don't think Zorch is doing that...
- I can make a loopback connection to myself from port to port and bring up
- SLIP over that wire, and it works just fine. It even trims some 200ms of
- latency off of pings...
-
- 2) If I watch what's happening from aforementioned nearby Sun, using
- etherfind, and try to telnet into another nearby Sun, I can see the SYNs
- and acks fly by, and some data packets and whatnot. If I then terminate
- the telnet I can see FINs and RESETs go by, and netstat indicates that
- the connection goes into FIN_WAIT_2 and then disappears properly.
-
- 3) I thought it was a data transparency problem. So, I discovered that
- ping has this nice facility of padding out packets with byte sequences,
- and if you ping with large enough packets it will go from 0x00 to 0xff
- in increasing order. Just lovely, only problem is that it showed pretty
- conclusively that nothing was munging bytes...
-
- My best guess is that it's a TCP header compression issue. I think that
- the Annex supports VJ compression, and from the 84-86 timeframe of the
- dates on my if_sl.c, I gather that Zorch doesn't. I thought that this
- was an optionally negotiated type of thing, and if one end didn't know
- about it things would still work OK, though less than optimally. Am I
- wrong? If so, how do I go about finding the right modules to update my
- driver and possibly associated header files?
-
- Thanks for any assistance.
-
- --
- Scott Hazen Mueller +1 408 285 5762 scott@dsg.tandem.com
- Tandem Computers, Unix System Administrator, Email Postmaster and Usenet Admin
- --
- Scott Hazen Mueller +1 408 285 5762 scott@dsg.tandem.com
- Tandem Computers, Unix System Administrator, Email Postmaster and Usenet Admin
-