home *** CD-ROM | disk | FTP | other *** search
- Newsgroups: comp.sys.sgi
- Path: sparky!uunet!cis.ohio-state.edu!zaphod.mps.ohio-state.edu!usc!rpi!utcsri!geac!alias!chk
- From: chk@alias.com (C. Harald Koch)
- Subject: Incoming UUCP flow control problem?
- Message-ID: <1992Sep1.152548.19218@alias.com>
- Sender: news@alias.com (News Owner)
- Organization: Alias Research, Inc., Toronto ON Canada
- Date: Tue, 1 Sep 1992 15:25:48 GMT
- Lines: 47
-
- We have a SGI IRIS-4D/80S running IRIX 4.0.1 with a 6-port serial board.
- There are two modems used exclusively for UUCP connected to the on-board
- modem ports, and four more modems connected to the 6-port board. All modems
- are Telebits; the two UUCP modems and two of the others are TB+; the
- remaining two are T2500s.
-
- We have several PEP UUCP links to various local and LD sites. We have a
- couple of V.32 links also, using the T2500s.
-
- The modems are all connected with "proper" cables; the two on-board ports
- use the standard flow-control cable described in the owners manual, and the
- CDSIO modems use a cable described to me by the TAC (pins 1 and 6
- disconnected). I'm using the ttyf device for all modems. The modems are all
- configured to use RTS/CTS full duplex handshaking, by the fix-telebit script.
-
- Problem:
-
- I've been seeing lousy throughput on all incoming UUCP connections, much
- worse than would be expected from even a busy 4D/80. (My 8MHz 68000 based
- Amiga can keep up with a TB+; I'd expect a 16MHz R2000 to have no problems).
- Throughputs have been 100-250 cps incoming, v.s. 1300+ outgoing on the PEP
- links; 100+ incoming v.s. 850+ outgoing on the V.32 links.
-
- I decided to turn UUCP debugging on, to see what's going on. I discovered
- the following: incoming UUCP data is being trashed and/or dropped on a
- regular basis. I'll see a normal incoming packet stream, and then something
- like ..., 1, 2, 3, 4, 6. Packet 4 is usually trashed, and packet 5 is
- missing altogether. A UUCP retry always fixes this, but the error eventually
- happens again. On large file transfers, the remote UUCP usually decides it's
- seen too many errors and hangs up.
-
- This looks suspiciously like a flow-control problem. It goes away altogether
- at 2400bps. It's worse on the on-board ports than it is on the 6-port board.
-
- I've also tried the same configuration files and cables on a 4D/35S running
- 4.0.5A. There were 0 errors in the transmission of a 940K file, and the
- throughput was 1401 cps.
-
- Help! I can't guess why this would be happening; to the best of my
- knowledge, everything is hooked up and configured correctly. Any suggestions
- from Out There would be greatly appreciated!
-
- --
- "You are at wit's end. | C. Harald Koch Alias Research, Inc. Toronto, ON
- Passages lead off in all | chk@alias.com (work-related mail)
- directions..." -related by | chk@gpu.utcs.utoronto.ca (permanent address)
- Bryan Manske, 13-Aug-92 | VE3TLA@VE3OY.#SCON.ON.CA.NA (AMPRNet)
-