On Mon, Jun 02, 2008 at 09:38:29AM -0400, Tom Lane wrote:
> That's not very surprising at all: a backend might have to write out a
> dirty buffer in order to reclaim the buffer for re-use, and which
> database the page is from doesn't enter into that.
> What does seem surprising is that it's had to do that quite a lot.
> Is this an old PG version without a bgwriter process? In a properly
> functioning recent release, I'd expect only the bgwriter to have a
> really eclectic mix of open files.
# select version();
version
-----------------------------------------------------------------------------------------------------------------------
PostgreSQL 8.2.7 on x86_64-unknown-linux-gnu, compiled by GCC gcc (GCC) 4.1.2 20061115 (prerelease) (Debian 4.1.1-21)
(1 row)
# select name, setting from pg_settings where name ~ 'bgw';
name | setting
-----------------------+---------
bgwriter_all_maxpages | 5
bgwriter_all_percent | 0.333
bgwriter_delay | 200
bgwriter_lru_maxpages | 5
bgwriter_lru_percent | 1
(5 rows)
should i show any more information?
regards,
depesz
