On Tue, Mar 04, 2008 at 09:56:13PM -0800, Paul Menage wrote:
quoted text
> Hi Kazunaga,
> 
> On Tue, Mar 4, 2008 at 9:39 PM, Kazunaga Ikeno <k-ikeno@ak.jp.nec.com> wrote:
> > Hi -
> >
> >  This is a patch of a new optional function for task assignment to cgroup, RFC.
> >
> >
> >  == Purpose =================================================
> >
> >  To provide the function that leads a task, corresponding to the conditions specified
> >  beforehand, to a specific cgroup directory.
> >
> 
> This is something that's been discussed before, originally as part of
> CKRM with a complex rule engine in the kernel space.
> 
> Basically, the general agreement was that it's a case where a simple
> API is going to be too simple for the majority of users, and a complex
> API that satisfies everyone is going to be too messy/heavyweight.
> 
> This is something that can be done in a userspace daemon via the
> process events connector - when you get a PROC_EVENT_UID event, you
> can move the process into the appropriate cgroup (you may also need to
> check any recently-forked children). This also gives you more
> flexibility than you can have in the kernel - you can base your
> decision on more complex factors than simply the uid of the process.
> 
> Dhaval Giani had a prototype implementation of such a daemon.
> 

The daemon was posted at
http://article.gmane.org/gmane.linux.kernel/553267 . At that point
control groups were called containers. These corrections will have to
made for it to run.

If I can get the time, I will clean it up and try to put it up
somewhere.

Thanks,

-- 
regards,
Dhaval
--
unsubscribe notice
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/