Lustre discuss - Nov 2006 - debugging lustre

Hi

I''m running lustre 1.6 beta 5 on i686 CentoOS 4 oss''s. I have
lately
experienced a kernel crash.
What can be done in order to debug it on the same platform (kernel 2.6.9
EL-smp).

I also try to cmompile lustre 1.6 beta 5 from code on FC5 2.6.15 kerenl
as I saw that you have provided  a config file for it in
"../kernel_configs/kernel-2.6.15-2.6-fc5-i686-smp.config".
However, I fail to apply the patches pointed
by ../series/2.6-fc5.series. I guess that the series file is not being
up to-date.

Any help about either is mostly welcome.

On Mon, 13 Nov 2006, Itamar Ofek wrote:
> Hi
>
> I''m running lustre 1.6 beta 5 on i686 CentoOS 4 oss''s. I
have lately
> experienced a kernel crash.
> What can be done in order to debug it on the same platform (kernel 2.6.9
> EL-smp).
If there is a kernel stack trace, you could post it here; setting up a 
serial console is probably a good idea.

Also, if you have some time, setting up a crash dump (LKCD or kdump) could 
be even better.

There are pointers here:
   https://mail.clusterfs.com/wikis/lustre/LinuxDebugging


HTH

-- 
Jean-Marc Saffroy - jean-marc.saffroy@ext.bull.net