Jim C. Nasby wrote:
> On Wed, Jun 21, 2006 at 01:21:05PM -0300, jody brownell wrote:
>> Jun 21 13:04:04 vanquish postgres: [19-1] DEBUG: "target": removed 5645231 row versions in 106508 pages
>> Jun 21 13:04:04 vanquish postgres: [19-2] DETAIL: CPU 3.37s/1.23u sec elapsed 40.63 sec.
>> Jun 21 13:04:04 vanquish postgres: [20-1] DEBUG: "target": found 5645231 removable, 1296817 nonremovable row versions in 114701 pages
>> Jun 21 13:04:04 vanquish postgres: [20-2] DETAIL: 0 dead row versions cannot be removed yet.
> So the table contained 5.6M dead rows and 1.3M live rows.
> I think you should forget about having autovacuum keep this table
> in-check and add manual vacuum commands to your code. Autovac is
> intended to deal with 99% of use cases; this is pretty clearly in the 1%
> it can't handle.
Maybe your free space map is configured to small, can you watch out for
log messages telling to increase it?
Markus Schaber | Logical Tracking&Tracing International AG
Dipl. Inf. | Software Development GIS
Fight against software patents in EU! www.ffii.org www.nosoftwarepatents.org
In response to
pgsql-performance by date
|Next:||From: Tom Lane||Date: 2006-06-23 16:09:41|
|Subject: Re: Occupation bloc in pages of table |
|Previous:||From: Markus Schaber||Date: 2006-06-23 12:32:29|
|Subject: Re: Help tuning autovacuum - seeing lots of relationbloat|