[Mimedefang] mimedefang-multiplexor crash

[Kees Theunissen]

On Thu, 1 Sep 2011, Ghislain wrote:

>hi,
>
>    i have from time to time a mimedefang-multiplexor crash. The
>mimedefang process is still here but mimedefang-multiplexor is down.
>Each time we have a :
>
>  kernel: mimedefang[10760]: segfault at 130 ip 00007faef285c900 sp
>0000000042c200f0 error 4 in libmilter.so.1.0.1[7faef2851000+f000]
>
>(debian lenny 64 bits)
>
>  I wonder how we could debug this and have an automatic way to restart
>the system when it triggers. The startup script of debian does not have
>a "status" command i could use for this and using a dumb "mimedefang
>start" do not work as it see the mimedefang process and think that all
>is ok and not down so it does not start. Of course a restart could not
>be used as if it work then restarting will kill performances especialy
>every 30 minutes ^^.
>
>
>   Any idea on how to prevent this ?

As I wrote in my other reply in this thread I think this is the bug
reported at http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=527862 .
Message #47 in that thread mentions a patch that can be downloaded
from http://j-chkmail.ensmp.fr/libmilter .
The "debian-lenny/" directory at that site contains patched libmilter
debian packages for "Lenny".
I didn't test/use those packages as I'm running Debian Squeeze.


A quick and dirty workaround is inspecting the mail-logs and restarting
mimedefang in case of errors.
Something like:

   tail -n100 /var/log/mail.log | \
   grep -q 'Milter (mimedefang): to error state' && \
   /etc/init.d/mimedefang restart

frequently run by cron should do the job (not jet tested).
The regular expression in the grep command is based on my logs after
the last crash a few weeks ago.


Regards,

Kees.

-- 
Kees Theunissen
F.O.M.-Institute for Plasma Physics Rijnhuizen, Nieuwegein, Netherlands
E-mail: theuniss at rijnh.nl,  Tel: (+31|0)306096724,  Fax: (+31|0)306031204