Path: utzoo!utgpu!news-server.csri.toronto.edu!cs.utexas.edu!ut-emx!ibmchs!auschs!awdprime!sanders.austin.ibm.com!sanders
From: sanders@sanders.austin.ibm.com (Tony Sanders)
Newsgroups: comp.unix.aix
Subject: Re: Help!  Runaway syslogd...
Message-ID: <7215@awdprime.UUCP>
Date: 1 May 91 22:29:07 GMT
References: <1991May1.192652.12002@dartvax.dartmouth.edu>
Sender: news@awdprime.UUCP
Organization: IBM AWD, Austin
Lines: 21

paw@eleazar.dartmouth.edu (Pat Wilson) writes:
> For the last two days I've had /etc/syslogd go nuts and start forking 
> copies of itself (filling up the process table along the way).  I don't
The only syslogd fork()'s (other than to make itself a daemon) is when it
wants to do nonblocking I/O to a terminal in wallmsg().  However it
sets a 30 second timeout around the write().

Send it a SIGBUS and poke through the core file and see if you
can find out what tty it was trying to write to.  Perhaps
one of your terminals is in a funny state (doesn't explain
why the alarm() isn't working nor why syslog is trying to
send lots of messages, unless it's a messages saying that syslog
is broken :-).

Try changing your syslog.conf file to log all messages in
/tmp/syslog.debug (or whatever) and see if it's sending lots
of messages, perhaps the message will give you a clue where to
look next.

-- sanders@cactus.org
I am not an IBM representative, I speak only for myself.