mirror of https://lore.kernel.org/lkml/
 help / color / mirror / Atom feed
From: Andrew Morton <akpm@osdl.org>
To: linux-kernel@vger.kernel.org
Cc: Jens Axboe <axboe@suse.de>, Nick Piggin <nickpiggin@yahoo.com.au>
Subject: Re: 2.6.12-rc2-mm3
Date: Mon, 11 Apr 2005 22:00:13 -0700	[thread overview]
Message-ID: <20050411220013.23416d5f.akpm@osdl.org> (raw)
In-Reply-To: <20050411012532.58593bc1.akpm@osdl.org>

Andrew Morton <akpm@osdl.org> wrote:
>
> - The anticipatory I/O scheduler has always been fairly useless with SCSI
>    disks which perform tagged command queueing.  There's a patch here from Jens
>    which is designed to fix that up by constraining the number of requests
>    which we'll leave pending in the device.
> 
>    The depth currently defaults to 1.  Tunable in
>    /sys/block/hdX/queue/iosched/queue_depth
> 
>    This patch hasn't been performance tested at all yet.  If you think it is
>    misbehaving (the usual symptom is processes stuck in D state) then please
>    report it, then boot with `elevator=cfq' or `elevator=deadline' to work
>    around it.

So it turns out that patch was broken.  I've fixed it locally and the
results are good, but odd.

The machine is a 4GB x86_64 with aic79xx controllers and MAXTOR
ATLAS10K4_73WLS disks.  ext2 filesystem.

The workload is continuous pagecache writeback versus
read-lots-of-little-files:

	while true
	do
		dd if=/dev/zero of=/mnt/sdb2/x bs=40M count=100 conv=notrunc
	done

versus

	find /mnt/sdb2/linux-2.4.25 -type f | xargs cat > /dev/null

we measure how long the find+cat takes.

2.6.12-rc2, 	as,	tcq depth=2:		7.241 seconds
2.6.12-rc2, 	as,	tcq depth=64:		12.172 seconds
2.6.12-rc2+patch,as,	tcq depth=64:		7.199 seconds
2.6.12-rc2, 	cfq2,	tcq depth=64:		much more than 5 minutes
2.6.12-rc2, 	cfq3,	tcq depth=64:		much more than 5 minutes

So

- The effects of tcq on AS are much less disastrous than I thought they
  were.  Do I have the wrong workload?  Memory fails me.  Or did we fix the
  anticipatory scheduler?

- as-limit-queue-depth.patch fixes things right up anyway.  Seems to be
  doing the right thing.  

- CFQ is seriously, seriously read-starved on this workload.

CFQ2:

procs -----------memory---------- ---swap-- -----io---- --system-- ----cpu----
 r  b   swpd   free   buff  cache   si   so    bi    bo   in    cs us sy id wa
 2  3   1116  25504   4868 3854008    4    0     8 61976 1112   291  0  4 39 58
 0  4   1112  24992   4868 3855120    0  568     4 53804 1124   452  0  4 54 43
 0  4   1112  24032   4868 3856004    0    0     8 44652 1110   303  0  3 45 53
 0  2   1112  25912   4872 3854164    0    0     4 51108 1122   321  0  3 52 45
 2  3   1112  24312   4872 3855728    0    0    32 52240 1113   300  0  4 44 52
 1  3   1112  25728   4876 3854432    0    0    20 48128 1118   296  0  3 58 39
 0  2   1112  23872   4876 3856336    0    0     4 48136 1116   288  0  4 47 49
 0  4   1112  25856   4876 3854300    0    4    16 50260 1117   294  0  3 55 42

CFQ3:

procs -----------memory---------- ---swap-- -----io---- --system-- ----cpu----
 r  b   swpd   free   buff  cache   si   so    bi    bo   in    cs us sy id wa
 1  5   1008  25888   4204 3845820    0    0    12 50544 1119   116  0  3 49 48
 0  5   1008  24096   4204 3847520    0    0     8 51200 1112   110  0  3 49 48
 0  5   1008  25824   4204 3845820    0    0     8 54816 1117   120  0  4 49 48
 0  5   1008  25440   4204 3846160    0    0     8 52880 1113   115  0  3 49 48
 0  5   1008  25888   4208 3845748    0    0    16 51024 1121   116  0  3 49 48



  parent reply	other threads:[~2005-04-12  5:25 UTC|newest]

Thread overview: 77+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2005-04-11  8:25 2.6.12-rc2-mm3 Andrew Morton
2005-04-11  8:56 ` 2.6.12-rc2-mm3 J.A. Magallon
2005-04-11  9:43   ` 2.6.12-rc2-mm3 Andrew Morton
2005-04-11 21:59     ` 2.6.12-rc2-mm3 Borislav Petkov
2005-04-11 22:22       ` 2.6.12-rc2-mm3 Andrew Morton
2005-04-12  4:20         ` 2.6.12-rc2-mm3 Stas Sergeev
2005-04-12  4:27           ` 2.6.12-rc2-mm3 Andrew Morton
2005-04-12 19:37             ` [patch 0/3] 2.6.12-rc2-mm3 Stas Sergeev
2005-04-12 19:42             ` [patch 1/3]: move config option for BAD_SYSCALL_EXIT Stas Sergeev
2005-04-12 19:47             ` [patch 2/3]: entry.S trap return fixes Stas Sergeev
2005-04-13  2:09               ` Andrew Morton
2005-04-13  3:18                 ` Stas Sergeev
2005-04-12 19:54             ` [patch 3/3]: fix BAD_SYSCALL_EXIT lockup Stas Sergeev
2005-04-12 12:22           ` 2.6.12-rc2-mm3 Borislav Petkov
2005-04-11 10:34 ` 2.6.12-rc2-mm3 Jan Dittmer
2005-04-11 17:33 ` 2.6.12-rc2-mm3 Benoit Boissinot
2005-04-11 19:11 ` 2.6.12-rc2-mm3 Jindrich Makovicka
2005-04-12  0:22   ` 2.6.12-rc2-mm3 Andrew Morton
2005-04-12  8:21     ` 2.6.12-rc2-mm3 Jindrich Makovicka
2005-04-11 20:46 ` 2.6.12-rc2-mm3 Martin J. Bligh
2005-04-11 22:24   ` 2.6.12-rc2-mm3 Benoit Boissinot
2005-04-12 22:32     ` 2.6.12-rc2-mm3 Martin J. Bligh
2005-04-11 21:05 ` 2.6.12-rc2-mm3: CONFIG_MODULES=n MTD compile error Adrian Bunk
2005-04-12  1:18 ` 2.6.12-rc2-mm3 Juergen Kreileder
2005-04-12  2:09   ` 2.6.12-rc2-mm3 Benjamin Herrenschmidt
2005-04-12  3:26   ` 2.6.12-rc2-mm3 Benjamin Herrenschmidt
2005-04-12  4:42     ` 2.6.12-rc2-mm3 Juergen Kreileder
2005-04-12  6:34       ` 2.6.12-rc2-mm3 Benjamin Herrenschmidt
2005-04-12 18:08         ` 2.6.12-rc2-mm3 Juergen Kreileder
2005-04-12 22:40           ` 2.6.12-rc2-mm3 Benjamin Herrenschmidt
2005-04-13  1:44             ` 2.6.12-rc2-mm3 Juergen Kreileder
2005-04-15 18:23   ` 2.6.12-rc2-mm3 Juergen Kreileder
2005-04-15 23:23     ` 2.6.12-rc2-mm3 Benjamin Herrenschmidt
2005-04-17  8:40       ` 2.6.12-rc2-mm3 Juergen Kreileder
2005-04-24  0:01         ` 2.6.12-rc2-mm3 Andrew Morton
2005-04-24  1:59           ` 2.6.12-rc2-mm3 Juergen Kreileder
2005-04-24  2:15             ` 2.6.12-rc2-mm3 Benjamin Herrenschmidt
2005-04-24  3:14               ` 2.6.12-rc2-mm3 Juergen Kreileder
2005-04-24  4:25               ` 2.6.12-rc2-mm3 Juergen Kreileder
2005-04-24  9:53               ` 2.6.12-rc2-mm3 Oleg Nesterov
2005-04-24 23:11                 ` 2.6.12-rc2-mm3 Juergen Kreileder
2005-04-25  0:09                   ` 2.6.12-rc2-mm3 Benjamin Herrenschmidt
2005-05-03  6:29                 ` 2.6.12-rc2-mm3 Andrew Morton
2005-05-03 10:42                   ` 2.6.12-rc2-mm3 Oleg Nesterov
2005-04-12  5:00 ` Andrew Morton [this message]
2005-04-12  5:51   ` 2.6.12-rc2-mm3 Nick Piggin
2005-04-12  6:19     ` 2.6.12-rc2-mm3 Andrew Morton
2005-04-12  6:49       ` 2.6.12-rc2-mm3 Nick Piggin
2005-04-12  7:50         ` 2.6.12-rc2-mm3 Andrew Morton
2005-04-12 19:03           ` 2.6.12-rc2-mm3 Steven Pratt
2005-04-12 17:01       ` 2.6.12-rc2-mm3 Steven Pratt
2005-04-12  7:06   ` 2.6.12-rc2-mm3 Jens Axboe
2005-04-12 11:32 ` 2.6.12-rc2-mm3 Ed Tomlinson
2005-04-12 11:39   ` 2.6.12-rc2-mm3 Andrew Morton
2005-04-14  0:15     ` 2.6.12-rc2-mm3 Ed Tomlinson
2005-04-14  0:20       ` 2.6.12-rc2-mm3 Andrew Morton
2005-04-14  0:38         ` 2.6.12-rc2-mm3 Ed Tomlinson
2005-04-14  0:54           ` 2.6.12-rc2-mm3 Andrew Morton
2005-04-17 21:32 ` [-mm patch] fix "make mandocs" Adrian Bunk
2005-04-17 22:27 ` 2.6.12-rc2-mm3 Alexander Nyberg
2005-04-17 22:36 ` 2.6.12-rc2-mm3 Alexander Nyberg
2005-04-19  2:03 ` 2.6.12-rc2-mm3: hostap: do not #include .c files Adrian Bunk
2005-04-19  2:12   ` Jouni Malinen
2005-04-26  0:49 ` 2.6.12-rc2-mm3 Randy.Dunlap
2005-04-26  1:06   ` 2.6.12-rc2-mm3 Andrew Morton
2005-04-26  3:17   ` 2.6.12-rc2-mm3 Greg KH
2005-04-26 16:15     ` 2.6.12-rc2-mm3 Randy.Dunlap
2005-04-27 10:41 ` 2.6.12-rc2-mm3 Alexander Nyberg
2005-04-13 18:36 2.6.12-rc2-mm3 Stas Sergeev
2005-04-13 20:04 ` 2.6.12-rc2-mm3 Ingo Molnar
2005-04-14 16:38   ` 2.6.12-rc2-mm3 Stas Sergeev
2005-04-17 23:39 2.6.12-rc2-mm3 Mikael Pettersson
2005-04-18  9:56 ` 2.6.12-rc2-mm3 Alexander Nyberg
2005-04-18 11:05 ` 2.6.12-rc2-mm3 Alexander Nyberg
2005-04-18 11:14   ` 2.6.12-rc2-mm3 Arjan van de Ven
2005-04-18 14:10     ` 2.6.12-rc2-mm3 Alexander Nyberg
2005-04-18 22:27 2.6.12-rc2-mm3 Mikael Pettersson

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=20050411220013.23416d5f.akpm@osdl.org \
    --to=akpm@osdl.org \
    --cc=axboe@suse.de \
    --cc=linux-kernel@vger.kernel.org \
    --cc=nickpiggin@yahoo.com.au \
    /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

all inboxes | Powered by JetHome®