[RFC] [PATCH] Cgroup based OOM killer controller

[Nikanth Karthikesan]

On Thursday 22 January 2009 08:58:43 KAMEZAWA Hiroyuki wrote:
> On Wed, 21 Jan 2009 16:38:21 +0530
>
> Nikanth Karthikesan <knikanth at suse.de> wrote:
> > As Alan Cox suggested/wondered in this thread,
> > http://lkml.org/lkml/2009/1/12/235 , this is a container group based
> > approach to override the oom killer selection without losing all the
> > benefits of the current oom killer heuristics and oom_adj interface.
> >
> > It adds a tunable oom.victim to the oom cgroup. The oom killer will kill
> > the process using the usual badness value but only within the cgroup with
> > the maximum value for oom.victim before killing any process from a cgroup
> > with a lesser oom.victim number. Oom killing could be disabled by setting
> > oom.victim=0.
> >
> > Signed-off-by: Nikanth Karthikesan <knikanth at suse.de>
>
> Assume following
>   - the usar can tell "which process should be killed at first"
>
> What is the difference between oom_adj and this cgroup to users ?

It is next to impossible to specify the order among say 10 memory hogging 
tasks using oom_adj. Using this oom-controller users can specify the exact 
order.

> If oom_adj is hard to use, making it simpler is a good way, I think.
> rather than adding new complication.
>
> It seems both of oom_adj and this cgroup will be hard-to-use functions
> for usual system administrators. But no better idea than using memcg
> and committing memory usage.
>

To use oom_adj effectively one should continuously monitor oom_score of all 
the processes, which is a complex moving target and keep on adjusting the 
oom_adj of many tasks which still cannot guarantee the order. This controller 
is deterministic and hence easier to use.

Thanks
Nikanth