All of lore.kernel.org
 help / color / mirror / Atom feed
From: Craig Shelley <craig@microtron.org.uk>
To: reiserfs-list@namesys.com
Subject: Re: More Slowdown - testscript
Date: Wed, 23 Nov 2005 01:31:36 +0000	[thread overview]
Message-ID: <1132709496.7650.24.camel@teratron.lan.etheus.net> (raw)
In-Reply-To: <3aa654a40511221127x68d059c5hfdfd4254aec0e1b5@mail.gmail.com>


[-- Attachment #1.1: Type: text/plain, Size: 1903 bytes --]

Hi,

The slow sync problems still exist with fresh vanilla kernel patched
with Reiser4 only (reiser4-for-2.6.14-1.patch). The slow sync occurs
regardless of which IO scheduler is in use.

craig@teratron.lan.etheus.net:~$ uname -a
Linux teratron.lan.etheus.net 2.6.14.2 #2 PREEMPT Tue Nov 22 23:48:39
GMT 2005 i686 GNU/Linux

See the attached testscript.sh 
These are the results from the script:

sync a1

real    0m1.301s
user    0m0.000s
sys     0m0.000s


sync a2

real    0m0.341s
user    0m0.004s
sys     0m0.000s


sync a3

real    0m0.341s
user    0m0.004s
sys     0m0.004s


sync a4

real    0m0.307s
user    0m0.000s
sys     0m0.000s


Performing recursive ls

real    0m0.307s
user    0m0.080s
sys     0m0.220s


sync b1

real    0m9.716s
user    0m0.000s
sys     0m0.024s


sync b2

real    0m0.391s
user    0m0.004s
sys     0m0.000s


sync b3

real    0m0.316s
user    0m0.000s
sys     0m0.000s


sync b4

real    0m0.341s
user    0m0.000s
sys     0m0.000s


Performing recursive ls

real    0m0.734s
user    0m0.216s
sys     0m0.516s


sync c1

real    0m53.698s
user    0m0.000s
sys     0m0.108s


sync c2

real    0m0.665s
user    0m0.000s
sys     0m0.004s


sync c3

real    0m0.125s
user    0m0.000s
sys     0m0.008s


sync c4

real    0m0.125s
user    0m0.000s
sys     0m0.000s




Sync a1->a4 execute relatively quickly
The recursive ls happen very quickly because the data is already cached.
Syncs immediately after the recursive ls take ages. This is really
strange since the recursive ls does not touch the disk because the data
is cached.
My guess is that when sync is called, the cache of ALL recently accessed
data is either being committed back to disk, or being re-read. 

-- 
Craig Shelley
EMail: craig@microtron.org.uk
Jabber: shell@jabber.earth.li

[-- Attachment #1.2: testscript.sh --]
[-- Type: application/x-shellscript, Size: 542 bytes --]

[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 189 bytes --]

  reply	other threads:[~2005-11-23  1:31 UTC|newest]

Thread overview: 50+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2005-11-17 13:47 More Slowdown Hesse, Christian
2005-11-17 17:22 ` More Slowdown or reiser4 update for 2.6.14-mm2 Vladimir V. Saveliev
2005-11-17 19:13   ` Thorsten Hirsch
2005-11-17 19:33     ` Thorsten Hirsch
2005-11-17 20:23       ` Thorsten Hirsch
2005-11-17 19:40   ` Hesse, Christian
2005-11-21 18:06     ` E.Gryaznova
2005-11-21 18:01       ` Artur Makówka
2005-11-21 18:44       ` Hesse, Christian
2005-11-21 20:50         ` Hesse, Christian
2005-11-21 22:51           ` Craig Shelley
2005-11-21 19:41       ` Avuton Olrich
2005-11-21 21:01       ` Sander
2005-11-22 18:46         ` E.Gryaznova
2005-11-22 19:27           ` Avuton Olrich
2005-11-23  1:31             ` Craig Shelley [this message]
2005-11-23  1:47               ` More Slowdown - testscript [Part 2] Craig Shelley
2005-11-23 20:23                 ` More Slowdown - testscript [noatime,nodiratime] Craig Shelley
2005-11-24  4:49                   ` Hans Reiser
2005-11-24  7:49                     ` Sander
2005-11-24  9:41                       ` Avuton Olrich
2005-11-24 10:13                         ` Artur Makówka
2005-11-23  7:09           ` Collect data? (was: Re: More Slowdown or reiser4 update for 2.6.14-mm2) Sander
2005-11-23 10:06             ` Collect data? E.Gryaznova
2005-11-23 10:15               ` Sander
2005-11-23 10:26             ` Collect data? (was: Re: More Slowdown or reiser4 update for 2.6.14-mm2) Hesse, Christian
2005-11-23 10:48               ` Sander
2005-11-23 11:22                 ` Collect data? Vladimir V. Saveliev
2005-11-23 11:53                   ` Hesse, Christian
2005-11-23 12:17                     ` Sander
2005-11-23 12:21                       ` Hesse, Christian
2005-11-23 12:28                         ` Sander
2005-11-23 13:09                           ` Vladimir V. Saveliev
2005-11-23 15:10                             ` Sander
2005-11-23 15:19                               ` Vladimir V. Saveliev
2005-11-23 15:31                                 ` Sander
2005-11-23 15:50                                   ` Vladimir V. Saveliev
2005-11-23 15:57                                     ` Sander
2005-11-23 18:42                                     ` Ingo Bormuth
2005-11-23 12:08                   ` Sander
2005-11-22 10:23 ` More Slowdown Marcel Hilzinger
2005-11-22 10:38   ` Sander
2005-11-22 10:44     ` Marcel Hilzinger
2005-11-22 14:29       ` David Masover
2005-11-22 15:01         ` Marcel Hilzinger
2005-11-22 19:15           ` David Masover
2005-11-22 23:17         ` Hans Reiser
2005-11-22 15:08       ` Sander
2005-11-22 11:34   ` Ingo Bormuth
2005-11-22 12:29     ` Artur Makówka

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=1132709496.7650.24.camel@teratron.lan.etheus.net \
    --to=craig@microtron.org.uk \
    --cc=reiserfs-list@namesys.com \
    /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 an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.