Re: [exim] exim dies on the interrupted system call

Top Page

Reply to this message
Author: W B Hacker
To: exim users
Subject: Re: [exim] exim dies on the interrupted system call
Артем Каялайнен wrote:
> Hi folks,
> I can't understand how to deal with the dying exim. It dies with this in
> panic log:
> 2010-12-25 23:17:44 [22683] failed to write to main log: length=115
> result=-1 errno=4 (Interrupted system call)
> I suppose that this is happening at the moment a high load on the file
> system (backup of mailboxes). At least this was never on weekdays, only on
> weekends, after or during the backup process. Other daemons running without
> problems.
> OS is FreeBSD 8.1-RELEASE, file system is UFS2 + gjournal (1Tb of free
> space) on raid5 with adaptec 3805 controller. Exim version 4.72
> Any suggestions?
> Thanks in advance.
> P.S. Sorry for my terrible english, i used online translator.
> ---
> WBR, Артём Каялайнен.

Should not be happening.

Generally on a 'vanilla' FreeBSD:

- your log is in /var/log/exim/[mainlog|rejectlog|paniclog]

- a 'conventional' mailstore may also be on the /var/ mount-point.

- mbox format can be more challenging w/r b/u than maildir. Which have you?

But none of the above should significantly affect Exim's ability to *log*, even
if the /var/ mountpoint is being heavily accessed. UFS2, the VMFS, and your RAID
controller should be able to juggle all that.


What cron reports hat look at the log, or what log rotations are being made at
the time of the error?

Is it possible Exim has prepped to do a write and had the 'current' logfile
rotated out from under it?