On Tue, Jul 13, 2010 at 01:51:54PM +0200, Claus Herwig wrote:
> Hello Herbert,
> >> Anybody knows whats happening? I assume that the host and even other
> >> guests should not be affected by a mad running guest....
> >
> > known problem with this kernel, enable memory cgroups
> > and use that (probably requires an update to util-vserver)
> > or avoid setting a hard limit on guests
> Could you please elaborate a bit on that "known problem"?
> Or provide a link to some kernel bugreport or anything?
the OOM killing inside a guest works the very same than
on the host, except for the fact that the selection is
limited to guest only processes ...
the problem with this approach is that under certain
circumstances (due to several changes in mainline) there
is no process to be killed inside a guest (init is protected
as on the host), and the OOM killer goes around in circles
(which sooner or later ends up with a panic)
that is why we dropped the restriction and memory limites
on recent kernels and replaced them with the cgroup limits
(which handle this case better on recent kernels)
HTC,
Herbert
> I've got one vserver host here which keeps dropping dead every
> 2-3 month without any known reason.
> And I know that before the last freeze it did a lot of OOM
> killing...
> Thank you,
> Claus
>
> --
> CHECON EDV-Consulting und Redaktion
> Claus Herwig * Barer Straße 70 * 80799 München
> +49 89 27826981 * Fax 27826982 * c.herwig@checon.de
Received on Tue Jul 13 17:20:03 2010