Re: [exim] Exim SIGSEGV

On 2009-04-03 at 12:38 +0400, Dmitry Samersoff wrote:
> Time to time I received messages:
>
> kernel: pid 16670 (exim-4.69-3), uid 27: exited on signal 11
> kernel: pid 16671 (exim-4.69-3), uid 27: exited on signal 11
> kernel: pid 16672 (exim-4.69-3), uid 27: exited on signal 11

This is ... rare.

> where I can get more information what happens?

In the main section of your config, you can have a "log_selector" option
to adjust what gets logged by Exim.

log_selector = +pid

After that, log-lines will include the pid after the time; eg, your
message was received by my system in pid 63580:

2009-04-03 08:40:45 [63580] 1LpexB-000GXU-18
<= exim-users-bounces+spodhuis.org@???
H=tahini.csx.cam.ac.uk [2001:630:200:8080:204:23ff:fed6:b664]:48881
I=[2a02:898:31:0:48:4558:736d:7470]:25 P=esmtp S=2626
id=49D5CAF6.6000902@??? for exim-users@???
2009-04-03 08:40:46 [63580]
SMTP connection from tahini.csx.cam.ac.uk [2001:630:200:8080:204:23ff:fed6:b664]:48881
I=[2a02:898:31:0:48:4558:736d:7470]:25 closed by QUIT

If you track it down enough to post for more help, please include
the output of:
exim -bV
ldd `which exim`
uname -a

Note that Exim will dump core if the OS will permit it to, except during
deliveries. You'll need to make sure the ulimit on cores is unlimited
and that the OS permits setuid files to coredump. Eg, on FreeBSD:
sysctl kern.sugid_coredump=1
and you can also force the core-dumps into a directory of your choice
with something like this (example from core(5)):
sysctl kern.corefile="/var/coredumps/%U/%N.core"
If you track the coredump down to happening during delivery,

-Phil