On Tuesday 08 February 2005 05:34, Dave Chinner wrote:
> On Tue, Feb 08, 2005 at 01:37:16AM +0100, Anders Saaby wrote:
>
> Uselessly chasing long chains in the hash is probably where the CPU
> usage is coming from, then.
Sounds right when hashsize is small - Looking at the code it seems default was
around 512 (?).
>
> > - If that is the case, do you have any ideas to how I can get my system
> > to perform better? (Performance gets very poor efter approx. two hours of
> > heavy load.)
>
> Seeing as you are running 2.6.11-rc3, you're in luck. Nathan's
> inode hash tweaks were merged into -rc3, so you can increase the
> size of the inode hash with a mount option. The mount option
> is "ihashsize", and it specifies the width of the hash table.
> i.e. `mount -o ihashsize=xxxxx /dev/sdb1 /mntpt`
I just tried:
mount -o ihashsize=65536,logdev=/dev/sdc1
Systime went from ~90% to ~5% !! :-)
This is so sweet. The server now performs amazing under any load I can throw
at it. - Only thing yet to show in the long run is stability ;)
> I'm not sure what the default your filesystem will be using, but
> a maximum of 16*PAGE_SIZE (=64k chains on i386/x86-64) applies to the
> default setting.
I couldn't really find out what hashsize was default before, so 64K was just a
(good?) guess. As I can not see any large increase in memory usage maybe
default values should be changed?
- Thank you for your help! - You guys have now provided me with invaluable
help more than once! :)
--
Med venlig hilsen - Best regards - Meilleures salutations
Anders Saaby
Systems Engineer
------------------------------------------------
Cohaesio A/S - Maglebjergvej 5D - DK-2800 Lyngby
Phone: +45 45 880 888 - Fax: +45 45 880 777
Mail: as@xxxxxxxxxxxx - http://www.cohaesio.com
------------------------------------------------
|