Re: stats collector suddenly causing lots of IO

2010/4/13 Chris <lists(at)deksai(dot)com>:
> I have a lot of centos servers which are running postgres.  Postgres isn't used
> that heavily on any of them, but lately, the stats collector process keeps
> causing tons of IO load.  It seems to happen only on servers with centos 5.
> The versions of postgres that are running are:
>
> 8.1.18
> 8.2.6
> 8.3.1
> 8.3.5
> 8.3.6
> 8.3.7
> 8.3.8
> 8.3.9
> 8.4.2
> 8.4.3
>
> I've tried turning off everything under RUNTIME STATISTICS in postgresql.conf
> except track_counts (since auto vacuum says it needs it), but it seems to have
> little affect on the IO caused by the stats collector.
>
> Has anyone else noticed this?  Have there been recent kernel changes
> that could cause this that anyone knows about?  Since we haven't touched
> postgres on these boxes since they were setup initially, I'm a bit baffled as
> to what might be causing the problem, and why I can't make it go away short of
> kill -STOP.
>
> Any suggestions would be much appreciated!

stats file is writed to disk every 500ms (can be change while building
postgres) but it have been improved in 8.4 and should be write only if
needed.

In 8.4 you can change the directory where to write the stat file with
the config param : stats_temp_directory Perhaps have a test and
change the filesystem (you might want to try a ramdisk and another fs
- ext3 -XFS-ext4 depending of your kernel) and see if it does change
something in your IO load.

Anyway it looks like it is centos 5 relative so what is your curernt
running kernel ? (and what FS )

>
>
> --
> Sent via pgsql-performance mailing list (pgsql-performance(at)postgresql(dot)org)
> To make changes to your subscription:
> http://www.postgresql.org/mailpref/pgsql-performance
>

--
Cédric Villemain