From owner-cvs-usrsbin  Fri Jan  3 08:10:57 1997
Return-Path: <owner-cvs-usrsbin>
Received: (from root@localhost)
          by freefall.freebsd.org (8.8.4/8.8.4) id IAA29446
          for cvs-usrsbin-outgoing; Fri, 3 Jan 1997 08:10:57 -0800 (PST)
Received: from spinner.DIALix.COM (root@spinner.DIALix.COM [192.203.228.67])
          by freefall.freebsd.org (8.8.4/8.8.4) with ESMTP id IAA29437;
          Fri, 3 Jan 1997 08:10:47 -0800 (PST)
Received: from spinner.DIALix.COM (peter@localhost.DIALix.oz.au [127.0.0.1])
          by spinner.DIALix.COM (8.8.4/8.8.4) with ESMTP id AAA25840;
          Sat, 4 Jan 1997 00:10:17 +0800 (WST)
Message-Id: <199701031610.AAA25840@spinner.DIALix.COM>
X-Mailer: exmh version 2.0beta 12/23/96
To: "Jordan K. Hubbard" <jkh@time.cdrom.com>
cc: joerg_wunsch@uriah.heep.sax.de (Joerg Wunsch),
        jkh@freefall.freebsd.org (Jordan K. Hubbard),
        CVS-committers@freefall.freebsd.org, cvs-all@freefall.freebsd.org,
        cvs-usrsbin@freefall.freebsd.org
Subject: Re: cvs commit: src/usr.sbin/syslogd syslogd.c 
In-reply-to: Your message of "Fri, 03 Jan 1997 07:48:00 PST."
             <23460.852306480@time.cdrom.com> 
Date: Sat, 04 Jan 1997 00:10:17 +0800
From: Peter Wemm <peter@spinner.DIALix.COM>
Sender: owner-cvs-usrsbin@FreeBSD.ORG
X-Loop: FreeBSD.org
Precedence: bulk

"Jordan K. Hubbard" wrote:
> I'll simply back this out of -current and 2.2 in the next few hours.
> No harm done, sorry - I wasn't aware of the savecore dependencies
> (how gross! :).

I've sent mail on this off list, but for what it's worth, in /etc/rc we 
have:

rm -f /var/run/log
echo ' syslogd.';                       syslogd

Apart from the possibility of bugs in the unlink() code (Bruce has 
commented on it), the change Jordan made should be about the same as 
moving the rm -f into syslogd itself.  It shouldn't make any difference.. 
(But then again, I've had a rough few weeks, I am probably missing 
something obvious again)

I'm also not sure why it was possible for savecore to do a send() to the 
/dev/log unix domain socket before anything had bound to it and started 
listening for datagrams on it.  Is this a special feature of AF_UNIX 
sockets?

> If the syslogd death problem were easy to catch, we'd probably have
> it fixed by now tho.. :(

Does anybody have more details on this?  I'd like to find out what's going 
on since I've got a lot of syslogd's blood on my hands over the last few 
months.

Do the people that are seeing this problem have user broadcasts in their 
/etc/syslog.conf file?  eg:
*.err                                           root
*.notice;news.err                               root
*.alert                                         root

Do the problems seem to happen around the time of a message coming through 
here?

[..]
> > >   Unlink LogName before attempting to open it.  Since we now have a lovel
    y
> > >   bug in syslogd which causes it to die after random amounts of time (wid
    el
> y
> > >   reported), this at least allows the administrator to easily restart it
> > >   without wondering why it simply exits again each time.

Cheers,
-Peter