From: | Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> |
---|---|
To: | Daryl Herzmann <akrherz(at)iastate(dot)edu> |
Cc: | pgsql-performance(at)postgresql(dot)org |
Subject: | Re: Missing the point of autovacuum |
Date: | 2007-01-06 19:46:40 |
Message-ID: | 17824.1168112800@sss.pgh.pa.us |
Views: | Raw Message | Whole Thread | Download mbox | Resend email |
Thread: | |
Lists: | pgsql-performance |
Daryl Herzmann <akrherz(at)iastate(dot)edu> writes:
> As the months have gone by, I notice many of my tables having *lots* of
> unused item pointers. For example,
> There were 31046438 unused item pointers.
> Total free space (including removable row versions) is 4894537260 bytes.
> 580240 pages are or will become empty, including 7 at the end of the
> table.
> 623736 pages containing 4893544876 free bytes are potential move
> destinations.
This definitely looks like autovac is not getting the job done for you.
The default autovac settings in 8.1 are very un-aggressive and many
people find that they need to change the settings to make autovac vacuum
more often. Combining autovac with the traditional approach of a cron
job isn't a bad idea, either, if you have known slow times of day ---
autovac doesn't currently have any concept of a maintenance window,
so if you'd rather your vacuuming mostly happened at 2AM, you still need
a cron job for that.
regards, tom lane
From | Date | Subject | |
---|---|---|---|
Next Message | Guy Rouillier | 2007-01-07 04:22:58 | Re: High update activity, PostgreSQL vs BigDBMS |
Previous Message | Tobias Brox | 2007-01-06 19:33:13 | Re: Missing the point of autovacuum |