Dell R900 with kernel panics while doing DB backups

Spike_White at Dell.com Spike_White at Dell.com
Thu Jan 12 10:57:52 CST 2012


Is this possibly a 32-bit kernel?  We used to get weekly kernel panics (due to low memory exhaustion) on our Linux DB servers.  But only during full DB backups.  

Of course, it wasn't clear at all that it was due to low memory exhaustion.   While circling the drain and basically unresponsive, Linux kernels can't always do in-depth diagnostics.  :-)

Like you, we did memtest, everything else.  All came back clean. In retrospect, a kdump would have helped I suppose.

Anyway, after we went to 64-bit kernels on all Linux DB servers yrs ago, we have not seen a recurrence of that problem.  

There's some tiny data structure movement you can do with some 32-bit kernel versions, move some structures from low kernel mem to high in order to relieve low mem pressure.  But that sounds like it'd be minimal help, a 64-bit kernel is the real answer.

Spike


----------------------------------------------------------------------

Message: 1
Date: Thu, 12 Jan 2012 14:22:54 -0200
From: "Eduardo Schoedler" <listas at esds.com.br>
Subject: Dell R900 with kernel panics
To: <linux-poweredge at dell.com>
Message-ID: <002201ccd146$7477b5a0$5d6720e0$@esds.com.br>
Content-Type: text/plain; charset="us-ascii"

Hi people,


This is my first email in this list.
Please forgive my bad english, I'm Brazilian.

I'm running one dedicated mysql server on Dell r900 with 2x Xeon E7310 (quad-core each) and 16GB RAM, OS Gentoo Linux amd64.

When I do backups (dumps) of the databases, some kerne panics occurs:

 

# cat /var/log/message | grep kernel
....
Jan  2 04:20:27 server kernel: [231912.188735] swapper: page allocation failure. order:1, mode:0x20 Jan  2 04:20:27 server kernel: [231912.188742] Pid: 0, comm: swapper Not tainted 2.6.39-gentoo-r4 #1 Jan  2 04:20:27 server kernel: [231912.188745] Call Trace:
Jan  3 04:11:16 server kernel: [317761.149866] swapper: page allocation failure. order:1, mode:0x20 Jan  3 04:11:16 server kernel: [317761.149873] Pid: 0, comm: swapper Not tainted 2.6.39-gentoo-r4 #1 Jan  3 04:11:16 server kernel: [317761.149876] Call Trace:
...

I don't know more what to do, memtest say that is no problem.
Any ideas?

Thanks in advance.

Best regards,

--
Eduardo Schoedler




More information about the Linux-PowerEdge mailing list