From: Mike Galbraith <efault@gmx.de>
To: Willy Tarreau <willy@w.ods.org>
Cc: lkml <linux-kernel@vger.kernel.org>, Ingo Molnar <mingo@elte.hu>,
Andrew Morton <akpm@osdl.org>, Con Kolivas <kernel@kolivas.org>
Subject: Re: interactive task starvation
Date: Tue, 21 Mar 2006 08:51:38 +0100 [thread overview]
Message-ID: <1142927498.7667.34.camel@homer> (raw)
In-Reply-To: <20060321064723.GH21493@w.ods.org>
[-- Attachment #1: Type: text/plain, Size: 1864 bytes --]
On Tue, 2006-03-21 at 07:47 +0100, Willy Tarreau wrote:
> Hi Mike,
Greetings!
> On Mon, Mar 20, 2006 at 08:09:13AM +0100, Mike Galbraith wrote:
> > real 0m0.512s
> > user 0m0.112s
> > sys 0m0.138s
> >
> > ...while running with the same apache loadavg of over 10, and tunables
> > set to server mode (0,0).
...
> Very good job !
> I told Grant in a private email that I felt confident the problem would
> quickly be solved now that someone familiar with the scheduler could
> reliably reproduce it. Your numbers look excellent, I'm willing to test.
> Could you remind us what kernel and what patches we need to apply to
> try the same, please ?
You bet. I'm most happy to have someone try it other than me :)
Apply the patches from the attached tarball in the obvious order to
2.6.16-rc6-mm2. As delivered, it's knobs are set up for a desktop box.
For a server, you'll probably want maximum starvation resistance, so
echo 0 > /proc/sys/kernel/grace_g1 and grace_g2. This will set the time
a task can exceed expected cpu (based upon sleep_avg) to zero seconds,
ie immediate throttling upon detection. It will also disable some
interactivity specific code in the scheduler.
If you want to fiddle with the knobs, grace_g1 is the number of CPU
seconds a new task is authorized to run completely free of any
intervention... startup in a desktop environment. grace_g2 is the
amount of CPU seconds a well behaved task can store for later usage.
With the throttling patch, an interactive task must earn the right to
exceed expected cpu by performing within expectations. The longer the
task behaves, the more 'good carma' it earns. This allows interactive
tasks to do a burst of activity, but the user determines how long that
burst==starvation is authorized. Tasks with just use as much cpu as
they can get run headlong into the throttle.
-Mike
[-- Attachment #2: throttle-V23-2.6.16-rc6-mm2.tar.gz --]
[-- Type: application/x-compressed-tar, Size: 7259 bytes --]
next prev parent reply other threads:[~2006-03-21 7:51 UTC|newest]
Thread overview: 112+ messages / expand[flat|nested] mbox.gz Atom feed top
2006-03-07 23:13 [PATCH] mm: yield during swap prefetching Con Kolivas
2006-03-07 23:26 ` Andrew Morton
2006-03-07 23:32 ` Con Kolivas
2006-03-08 0:05 ` Andrew Morton
2006-03-08 0:51 ` Con Kolivas
2006-03-08 1:11 ` Andrew Morton
2006-03-08 1:12 ` Con Kolivas
2006-03-08 1:19 ` Con Kolivas
2006-03-08 1:23 ` Andrew Morton
2006-03-08 1:28 ` Con Kolivas
2006-03-08 2:08 ` Lee Revell
2006-03-08 2:12 ` Con Kolivas
2006-03-08 2:18 ` Lee Revell
2006-03-08 2:22 ` Con Kolivas
2006-03-08 2:27 ` Lee Revell
2006-03-08 2:30 ` Con Kolivas
2006-03-08 2:52 ` [ck] " André Goddard Rosa
2006-03-08 3:03 ` Lee Revell
2006-03-08 3:05 ` Con Kolivas
2006-03-08 21:07 ` Zan Lynx
2006-03-08 23:00 ` Con Kolivas
2006-03-08 23:48 ` Zan Lynx
2006-03-09 0:07 ` Con Kolivas
2006-03-09 3:13 ` Zan Lynx
2006-03-09 4:08 ` Con Kolivas
2006-03-09 4:54 ` Lee Revell
2006-03-08 7:51 ` Jan Knutar
2006-03-08 8:39 ` Con Kolivas
2006-03-09 8:57 ` Helge Hafting
2006-03-09 9:08 ` Con Kolivas
[not found] ` <4410AFD3.7090505@bigpond.net.au>
2006-03-10 9:01 ` [ck] " Andreas Mohr
2006-03-10 9:11 ` Con Kolivas
2006-03-08 22:24 ` Pavel Machek
2006-03-09 2:22 ` Nick Piggin
2006-03-09 2:30 ` Con Kolivas
2006-03-09 2:57 ` Nick Piggin
2006-03-09 9:11 ` Con Kolivas
2006-03-08 13:36 ` [ck] " Con Kolivas
2006-03-17 9:06 ` Ingo Molnar
2006-03-17 10:46 ` interactive task starvation Mike Galbraith
2006-03-17 17:15 ` Mike Galbraith
2006-03-20 7:09 ` Mike Galbraith
2006-03-20 10:22 ` Ingo Molnar
2006-03-21 6:47 ` Willy Tarreau
2006-03-21 7:51 ` Mike Galbraith [this message]
2006-03-21 9:13 ` Willy Tarreau
2006-03-21 9:14 ` Ingo Molnar
2006-03-21 11:15 ` Willy Tarreau
2006-03-21 11:18 ` Ingo Molnar
2006-03-21 11:53 ` Con Kolivas
2006-03-21 13:10 ` Mike Galbraith
2006-03-21 13:13 ` Con Kolivas
2006-03-21 13:33 ` Mike Galbraith
2006-03-21 13:37 ` Con Kolivas
2006-03-21 13:44 ` Willy Tarreau
2006-03-21 13:45 ` Con Kolivas
2006-03-21 14:01 ` Mike Galbraith
2006-03-21 14:17 ` Con Kolivas
2006-03-21 15:20 ` Con Kolivas
2006-03-21 17:50 ` Willy Tarreau
2006-03-22 4:18 ` Mike Galbraith
2006-03-21 17:51 ` Mike Galbraith
2006-03-21 13:38 ` Willy Tarreau
2006-03-21 13:48 ` Mike Galbraith
2006-03-21 12:07 ` Mike Galbraith
2006-03-21 12:59 ` Willy Tarreau
2006-03-21 13:24 ` Mike Galbraith
2006-03-21 13:53 ` Con Kolivas
2006-03-21 14:17 ` Mike Galbraith
2006-03-21 14:19 ` Con Kolivas
2006-03-21 14:25 ` Ingo Molnar
2006-03-21 14:28 ` Con Kolivas
2006-03-21 14:30 ` Ingo Molnar
2006-03-21 14:28 ` Mike Galbraith
2006-03-21 14:30 ` Con Kolivas
2006-03-21 14:32 ` Ingo Molnar
2006-03-21 14:44 ` Willy Tarreau
2006-03-21 14:52 ` Ingo Molnar
2006-03-29 3:01 ` Lee Revell
2006-03-29 5:56 ` Ray Lee
2006-03-29 6:16 ` Lee Revell
2006-03-21 14:36 ` Mike Galbraith
2006-03-21 14:39 ` Con Kolivas
2006-03-21 14:39 ` Willy Tarreau
2006-03-21 18:39 ` Rafael J. Wysocki
2006-03-21 19:32 ` Willy Tarreau
2006-03-21 21:47 ` Rafael J. Wysocki
2006-03-21 22:51 ` Peter Williams
2006-03-22 3:49 ` Mike Galbraith
2006-03-22 3:59 ` Peter Williams
2006-03-22 12:14 ` [interbench numbers] " Mike Galbraith
2006-03-22 20:27 ` Con Kolivas
2006-03-23 3:22 ` Mike Galbraith
2006-03-23 5:43 ` Con Kolivas
2006-03-23 5:53 ` Mike Galbraith
2006-03-23 11:07 ` Mike Galbraith
2006-03-24 0:21 ` Con Kolivas
2006-03-24 5:02 ` Mike Galbraith
2006-03-24 5:04 ` Con Kolivas
2006-03-17 12:38 ` [PATCH] sched: activate SCHED BATCH expired Con Kolivas
2006-03-17 13:07 ` Ingo Molnar
2006-03-17 13:26 ` Nick Piggin
2006-03-17 13:36 ` Con Kolivas
2006-03-17 13:46 ` Nick Piggin
2006-03-17 13:51 ` Nick Piggin
2006-03-17 14:11 ` Con Kolivas
2006-03-17 14:59 ` Ingo Molnar
2006-03-17 13:47 ` [ck] " Andreas Mohr
2006-03-17 13:59 ` Con Kolivas
2006-03-17 14:06 ` Nick Piggin
2006-03-08 8:48 ` [ck] Re: [PATCH] mm: yield during swap prefetching Andreas Mohr
2006-03-08 8:52 ` Con Kolivas
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=1142927498.7667.34.camel@homer \
--to=efault@gmx.de \
--cc=akpm@osdl.org \
--cc=kernel@kolivas.org \
--cc=linux-kernel@vger.kernel.org \
--cc=mingo@elte.hu \
--cc=willy@w.ods.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox