From: Matt Ayres (matta_at_kindhosts.com)
Date: Sun 08 Dec 2002 - 23:46:10 GMT

• Next message: hugh at atosc dot org: "[vserver] per vserver fairness (CPU)"
• Previous message: Herbert Poetzl: "Re: [vserver] O(1) Scheduler anyone?"
• In reply to: Herbert Poetzl: "Re: [vserver] O(1) Scheduler anyone?"

I've started doing some work on this. The way the O(1) scheduler works
the lack of the sched option wouldn't be that bad. I'm having some issues
with unistd.h and exit.c... after that compiles i'll have to see how it
works (aka. can one vserver kill the server).

I cc'ed Alexey on my original email to this list but have not heard
anything.

--
Thanks,
Matt
On Sun, 8 Dec 2002, Herbert Poetzl wrote:
> On Sun, Dec 08, 2002 at 01:11:04PM -0500, Matt Ayres wrote:
> > Well, i'm already aware of both of these and am already using a
> > pre-emptive kernel. Due to the changes it looks like the only way to do
> > this quickly is just to remove the 'sched' code from the patch.
>
> - are you interested in collaborating on this issue?
> - if so, maybe we should contact Alexey Lyashkov?
> - I would like to see per VS memory limits in CTX-15
>   (Alexey seems to have done that)
>
> best,
> Herbert
>
> > --
> > Thanks anyway,
> > Matt
> >
> >
> > On Sun, 8 Dec 2002, Herbert Poetzl wrote:
> >
> > > On Sun, Dec 08, 2002 at 03:27:48AM -0500, Matt Ayres wrote:
> > > > So i'm running into a situation here.  I have a server running quite a few
> > > > virtual servers with over 1000 processes on the host server.  I am
> > > > experiencing some performance problems, not so much on the virtual servers
> > > > but any type of cpu intensive operation on the host server causes severe
> > > > load.
> > >
> > > have a look at ...
> > >
> > > http://www.kernel.org/pub/linux/kernel/people/rml/sched/ingo-O1/
> > >
> > > but in your case, I would also take a look at ...
> > >
> > > http://www.tech9.net/rml/linux/
> > >
> > > best,
> > > Herbert
> > >
> > > > I know Lyashkov Alexey has ported the Vserver patch to work with RedHat
> > > > 7.3 and the O(1) scheduler.  Has anyone put any consideration or already
> > > > done work to make a vserver+O(1) patch available for the official kernel?
> > > >
> > > > I'm going to try to attempt this using Lyashkov's as a base however I
> > > > havn't used C in quite some years and have yet to do any kernel hacking :P
> > > >
> > > > --
> > > > Thanks,
> > > > Matt Ayres
> > >
>

• Next message: hugh at atosc dot org: "[vserver] per vserver fairness (CPU)"
• Previous message: Herbert Poetzl: "Re: [vserver] O(1) Scheduler anyone?"
• In reply to: Herbert Poetzl: "Re: [vserver] O(1) Scheduler anyone?"