lamy@ai.utoronto.ca (Jean-Francois Lamy) (12/20/89)
We are seeing several dozen "broken pipes" on jarvis.csri.toronto.edu (running
nntpd 1.5.7 with C news batching, SunOS 3.5 on a 3/180). Two machines in
particular seem to be causing those: a microVax on some other network and a
Sun 3/280 running 4.0.3, on the local ethernet. Their news admin has been able
to track down the problem to their nntplink being returned a rather weird
error message from our nntpd, namely the entry for the "news" user in the
passwd file, but without proper termination.
This sounds vaguely familiar. Has anyone tracked this down to some
misbehaviour in either nntplink or nntpd? We are also seeing such broken
pipes reported by a B news site, even though nothing seems to be
wrong on the feeding machine (yet another machine on the campus lan).
Jean-Francois Lamy lamy@ai.utoronto.ca, uunet!ai.utoronto.ca!lamy
AI Group, Department of Computer Science, University of Toronto, Canada M5S 1A4
------- Forwarded Message
Replied: Tue, 19 Dec 89 21:57:35 -0500
Replied: "Paul Kern <pkern@gpu.utcs.utoronto.ca> +outbox"
Return-Path: gpu.utcs.toronto.edu!pkern
Received: from gpu.utcs.utoronto.ca by neat.cs.toronto.edu with SMTP id 2247; Tue, 19 Dec 89 20:48:01 EST
Received: by gpu.utcs.utoronto.ca id 63965; Tue, 19 Dec 89 20:46:57 EST
From: Paul Kern <pkern@gpu.utcs.utoronto.ca>
To: lamy@neat.cs.toronto.edu
Subject: Re: nntp
Message-Id: <89Dec19.204657est.63965@gpu.utcs.utoronto.ca>
Date: Tue, 19 Dec 89 20:46:50 EST
Hi.
Here's what I've been able to track down so far.
With each failure sfgets() in readreply() in gpu's nntplink returns
"436 :*:161:26:News user id,,,,,O,:/usr/spool/bnews:\n"
with "436 " being the part we're supposed to get (ERR_XFERFAIL)
and I guess the rest is the result of some pointer wierdness on
the jarvis nntpd(?) (I assume the rest of the line would normally
contain an appropriate error message like "disk full" or something (?)).
(Our nntplink's first complaint was the line didn't end with \r\n :-).
(The next mistake was that it counted on errno to be set properly).
Hope this helps.
pk.
------- End of Forwarded Message