Server is debian lenny 64 bits, 32 GB, 4 CPU cores. Postfix package is
version 2.5.5-1.1
The above mentioned server has a problem that postfix keeps terminating
at regular intervals. I believe I have narrowed the problem down to a
program eating a lot of memory, likely due to multiple forks each
wanting to allocate 10 GB, and causing postfix to quit. I have no
control over that program's source code. The CPU utilisation is above
95% nearing 100% when this happens.
I am not sure if it's the kernel's OOM killer or postfix itself which
causes it to quit. I do not see a mention in /var/log/messages about the
OOM killer killing postfix, but I do see a mention in /var/log/mail.info
about reading "postfix/master[14959]: terminating on signal 15" which is
the default kill signal.
Would postfix kill itself if it failed to allocate memory? To me that
seems the preferred behaviour for any such program. Or what is postfix'
behaviour under such high load conditions? It's being run with -o stress.
Is it possible to find out if in fact the OOM killer killed it, lacking
a specific log entry indicating it?
Thank you,
Jeroen
