Re: performance; disk bad or something? - Mailing list pgsql-general

From Michael Fuhr
Subject Re: performance; disk bad or something?
Date
Msg-id 20070407084137.GA73067@winnie.fuhr.org
Whole thread Raw
In response to Re: performance; disk bad or something?  (Marcus Engene <mengpg2@engene.se>)
Responses Re: performance; disk bad or something?
List pgsql-general
On Sat, Apr 07, 2007 at 09:28:58AM +0200, Marcus Engene wrote:
> INFO:  "apa_item_common": removed 9028 row versions in 3651 pages
> DETAIL:  CPU 0.24s/0.36u sec elapsed 30.69 sec.
> INFO:  "apa_item_common": found 9028 removable, 12863 nonremovable row
> versions in 14489 pages
> DETAIL:  0 dead row versions cannot be removed yet.
> There were 76646 unused item pointers.

How often does this table receive updates and deletes and how often
are you vacuuming it?  It averages less than one row per page (12863
nonremovable row versions in 14489 pages) and appears to have become
quite bloated with dead rows sometime in the past (76646 unused
item pointers).  Use CLUSTER or VACUUM FULL + REINDEX to compact
the table and run ANALYZE afterwards to update the statistics, then
make sure you're vacuuming it often enough to keep it from becoming
bloated again.

Is your free space map sufficiently sized?  If you do a database-wide
VACUUM VERBOSE, what are the last few lines of the output that
mention free space map settings?

--
Michael Fuhr

pgsql-general by date:

Previous
From: Marcus Engene
Date:
Subject: Re: performance; disk bad or something?
Next
From: Martijn van Oosterhout
Date:
Subject: Re: New to concurrency