public inbox for gentoo-user@lists.gentoo.org
 help / color / mirror / Atom feed
From: Daniel <danny@ilievnet.com>
To: gentoo-user@lists.gentoo.org
Subject: Re: [gentoo-user]  linux' IO performance sucks
Date: Wed, 05 Jul 2006 13:43:18 +0300	[thread overview]
Message-ID: <44AB97C6.4040801@ilievnet.com> (raw)
In-Reply-To: <e8fci1$6ui$1@sea.gmane.org>

Sven Köhler wrote:
> Hi,
> 
> sorry for the silly subject, but did you ever experience the following?:
> 
> i have a fileserver, i copy a file to it - let's say 600MB.
> 
> So about every 200MB (i guess the linux box writes the data into the
> cache in the RAM first) linux writes the harddisk. But during that time
> - during the time it writes that 200MB to disk, there is no chance for
> any other IO. I'm playing an mp3 from the very same fileserver. It stops
> playing, because the machine does answer the read-requests.
> 
> So what's going on here?
> 
> Why does Linux write so huge amounts of data to the disk? Why does Linux
> not stop writing for a while to fullfil the read-requests? And so on ...
> 
> Any idea, on how to imrpove that?
> 

Perhaps a more often flush of buffers may help you in this situation.

There are several parameters you can tweak to control your kernel
behavior regarding this.

You can put the following lines in your /etc/sysctl.conf file, replacing
 "i","j","k" and "l" with proper numbers.
vm.dirty_expire_centisecs =  i
vm.dirty_writeback_centisecs = j
vm.dirty_ratio = k
vm.dirty_background_ratio = l

The meaning of these parameters is descibed in the kernel documentation:
/usr/src/linux/Documentation/filesystems/proc.txt
/usr/src/linux/Documentation/sysctl/vm.txt

You could also disable all write caching by issuing the command:

hdparm -W0 /dev/<your-physical-disk-name>


Hope This Helps

---
Best regards
Daniel
-- 
gentoo-user@gentoo.org mailing list



  parent reply	other threads:[~2006-07-05 10:59 UTC|newest]

Thread overview: 10+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2006-07-05  3:43 [gentoo-user] linux' IO performance sucks Sven Köhler
2006-07-05  4:31 ` Thomas Cort
2006-07-05  4:33 ` Richard Fish
2006-07-05  5:45 ` Hemmann, Volker Armin
2006-07-05 10:38 ` Daniel Drake
2006-07-05 10:43 ` Daniel [this message]
2006-07-05 21:55   ` Hemmann, Volker Armin
2006-07-05 22:51     ` Daniel Iliev
2006-07-06  0:45     ` Daniel Iliev
2006-07-06  2:01       ` Dale

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=44AB97C6.4040801@ilievnet.com \
    --to=danny@ilievnet.com \
    --cc=gentoo-user@lists.gentoo.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