From: Matt Ayres (matta_at_tektonic.net)
Date: Mon 17 Nov 2003 - 18:40:28 GMT
On Mon, 2003-11-17 at 12:45, Martin Josefsson wrote:
> On Mon, 2003-11-17 at 12:23, Matt Ayres wrote:
> I've seen this one on a few diffrent systems, even servers at kernel.org
> has seen this exact problem but noone has been able to reproduce it at
> will or trace it, the machines seem to "half-hang" without printing
> anything on the console. (this is without the vserver patch)
>
> Do you have a simple testcase that manages to reproduce it? That would
> be wonderful.
>
> sysrq-t (I think it's t that's interesting)
> and
> ctrl - scroll lock
> and
> altgr - scroll lock
>
> That data together with a kernel compiled with -g (debugging) would be
> wonderful when it's "half-hung".
I wish I had a test case... then i'd just do it on a test server here
:) I'll be able to get a sysrq as soon as I start running vserver under
UML. I would say this hang is caused by the workload, many processes
and heavy cache usage. I imagine kernel.org has the same type of
workload.
-- Matt Ayres <matta_at_tektonic.net> TekTonic_______________________________________________ Vserver mailing list Vserver_at_list.linux-vserver.org http://list.linux-vserver.org/mailman/listinfo/vserver