Hi,
OK I've figured out what the problem is now. I use Software
Suspend in my kernel, and /etc/suspend.conf was set to run
"mountnfs.sh" on resume. The new mountnfs.sh in Sarge runs
bootclean.sh (it didn't do that before), and bootclean.sh
cleans up /var/run. So /var/run was being cleaned every
time I resumed from suspend in the morning, even though
sysklogd was still running (it was resumed from suspend).
So when /etc/cron.daily/sysklogd tried to do
"/etc/init.d/sysklogd reload-or-restart" which invoked
start-stop-daemon to do its work, start-stop-daemon saw
there was no PID file but there WAS another running syslogd
process, and so it did NOTHING. My "/usr/bin/killall -HUP
syslogd" in /etc/logrotate.conf seemed to fix it becuase I
hadn't actually qualified it with syslog so it did it after
rotating every log (and logrotate runs just after
cron.daily/sysklogd, so it seemed to "fix" the problem).
So it's not a "logrotate" bug. However there may be an
initscripts bug to report, since I don't see why mountnfs.sh
should run bootclean.sh, given that an admin might well want
to use mountnfs.sh to mount the NFS independently of
cleaning /var/run. (I notice that there's already a bug
filed about this, bug 286479. I'll put some more info in it.)
Silas
Paul Martin writes:
> tag 316983 unreproducible moreinfo
> thanks
>
> On Tue, Jul 05, 2005 at 11:22:39AM +0100, Silas S. Brown wrote:
>
> > After I upgraded from woody to sarge, syslogd is no longer
> > signalled when the logs are rotated. The effect is that
> > syslogd continues to write to /var/log/syslog.0 instead of
> > the new /var/log/syslog, and my log tracking scripts (which
> > assume writes will always go to /var/log/syslog) fails to
> > catch the new entries.
>
> ...but logrotate isn't responsible for rotating syslogd's logs.
>
> /etc/cron.{daily,weekly,monthly}/sysklogd does it.
>
> That ends with...
>
> # Restart syslogd
> #
> /etc/init.d/sysklogd reload-or-restart > /dev/null
>
>
> --
> Paul Martin <[EMAIL PROTECTED]>
--
Silas S. Brown, Cambridge Univ. Computer Lab, http://www.cus.cam.ac.uk/~ssb22
"The rationality of our universe is best suggested by the fact that we
can discover more about it from any starting point, as if it were a
fabric that will unravel from any thread" - George Zebrowski, 1994
--
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
