Bacula-users

Re: [Bacula-users] bacula-sd dmesg output

2011-05-16 09:11:15
Subject: Re: [Bacula-users] bacula-sd dmesg output
From: Konstantin Khomoutov <flatworm AT users.sourceforge DOT net>
To: John Drescher <drescherjm AT gmail DOT com>
Date: Mon, 16 May 2011 17:08:01 +0400
On Mon, 16 May 2011 08:39:38 -0400
John Drescher <drescherjm AT gmail DOT com> wrote:

> > for a couple of weeks now I occasionally see messages like these in
> > the kernel log or our bacula server (bacula 5.0.3 compiled from
> > source on centos 5.6 64bit):
> >
> > ######################################################################
> > INFO: task bacula-sd:12467 blocked for more than 120 seconds.
> > "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this
[...]
> Maybe do what it says.
> 
> echo 0 > /proc/sys/kernel/hung_task_timeout_secs
I would rather not do this.
Tasks hanging on some kind of resource may signalize a problem with the
kernel or with the underlying hardware or the combination of the two.

But since we have almost zero amount of information about what kind of
activity Bacula performs (we just know SCSI layer is somehow involved
but this gives almost zero info on its own) it's hard to guess what
might be the cause of the problem.

Just one fact off the top of my memory: this syndrome of tasks being
hung has been present in bug reports regarding activity on certain file
systems under massive fsync()-ing.

------------------------------------------------------------------------------
Achieve unprecedented app performance and reliability
What every C/C++ and Fortran developer should know.
Learn how Intel has extended the reach of its next-generation tools
to help boost performance applications - inlcuding clusters.
http://p.sf.net/sfu/intel-dev2devmay
_______________________________________________
Bacula-users mailing list
Bacula-users AT lists.sourceforge DOT net
https://lists.sourceforge.net/lists/listinfo/bacula-users

<Prev in Thread] Current Thread [Next in Thread>