[Pkg-mailman-hackers] Bug#505638: mailman: Mailman dies on log rotation

[Robert Henney]

On Mon, Jan 16, 2012 at 05:36:04PM +0100, Ivan Marton wrote:
> Hi!
> 
> I've faced the same problem.

This bug took my mailman down this last Sunday.  In fact, you'll notice
that reports of this bug appear to indicate that it only occurs 
during Sunday morning cron execution.  I believe this is because 
the /etc/logrotate.d/mailman file is written to execute 
two "mailmanctl -q reopen" processes at the same time on Sundays.  (so 
this bug may also trigger on month boundaries, but weekly/daily 
concurrency happens more frequently)

I would not be at all surprised if the mystery SIGTERM is being generated
by one of the two mailmanctl processes, possibly (but not knowing python,
uncertainly) at the other mailmanctl instance or at itself, which is then 
being passed on to the other qrunner processes.

my qrunner log:

Oct 21 06:32:28 2012 (1788) Master watcher caught SIGHUP.  Re-opening log files.
Oct 21 06:32:29 2012 (1795) VirginRunner qrunner caught SIGHUP.  Reopening logs.
Oct 21 06:32:29 2012 (1795) VirginRunner qrunner caught SIGHUP.  Reopening logs.
Oct 21 06:32:29 2012 (1795) VirginRunner qrunner caught SIGTERM.  Stopping.
Oct 21 06:32:29 2012 (1791) CommandRunner qrunner caught SIGHUP.  Reopening logs.
Oct 21 06:32:29 2012 (1791) CommandRunner qrunner caught SIGTERM.  Stopping.
Oct 21 06:32:29 2012 (1791) CommandRunner qrunner caught SIGHUP.  Reopening logs.
Oct 21 06:32:29 2012 (1795) VirginRunner qrunner exiting.
Oct 21 06:32:29 2012 (1794) OutgoingRunner qrunner caught SIGHUP.  Reopening logs.
Oct 21 06:32:29 2012 (1791) CommandRunner qrunner exiting.
Oct 21 06:32:29 2012 (1794) OutgoingRunner qrunner caught SIGTERM.  Stopping.
Oct 21 06:32:29 2012 (1794) OutgoingRunner qrunner caught SIGHUP.  Reopening logs.
Oct 21 06:32:29 2012 (1793) NewsRunner qrunner caught SIGHUP.  Reopening logs.
Oct 21 06:32:29 2012 (1794) OutgoingRunner qrunner exiting.
Oct 21 06:32:29 2012 (1793) NewsRunner qrunner caught SIGTERM.  Stopping.
Oct 21 06:32:29 2012 (1792) IncomingRunner qrunner caught SIGHUP.  Reopening logs.
Oct 21 06:32:29 2012 (1793) NewsRunner qrunner caught SIGHUP.  Reopening logs.
Oct 21 06:32:30 2012 (1792) IncomingRunner qrunner caught SIGTERM.  Stopping.
Oct 21 06:32:30 2012 (1792) IncomingRunner qrunner caught SIGHUP.  Reopening logs.
Oct 21 06:32:30 2012 (1792) IncomingRunner qrunner exiting.
Oct 21 06:32:30 2012 (1793) NewsRunner qrunner exiting.
Oct 21 06:32:29 2012 (1789) ArchRunner qrunner caught SIGHUP.  Reopening logs.
Oct 21 06:32:30 2012 (1789) ArchRunner qrunner caught SIGTERM.  Stopping.
Oct 21 06:32:30 2012 (1789) ArchRunner qrunner caught SIGHUP.  Reopening logs.
Oct 21 06:32:30 2012 (1796) RetryRunner qrunner caught SIGHUP.  Reopening logs.
Oct 21 06:32:30 2012 (1789) ArchRunner qrunner exiting.
Oct 21 06:32:30 2012 (1796) RetryRunner qrunner caught SIGTERM.  Stopping.
Oct 21 06:32:30 2012 (1790) BounceRunner qrunner caught SIGHUP.  Reopening logs.
Oct 21 06:32:30 2012 (1796) RetryRunner qrunner caught SIGHUP.  Reopening logs.
Oct 21 06:32:30 2012 (1790) BounceRunner qrunner caught SIGTERM.  Stopping.
Oct 21 06:32:30 2012 (1790) BounceRunner qrunner caught SIGHUP.  Reopening logs.
Oct 21 06:32:31 2012 (1790) BounceRunner qrunner exiting.
Oct 21 06:32:31 2012 (1796) RetryRunner qrunner exiting.

This posting to Mailman-Users could be related:
http://www.mail-archive.com/mailman-users@python.org/msg60891.html