Re: vacuum and 24/7 uptime - Mailing list pgsql-general

From Philip Molter
Subject Re: vacuum and 24/7 uptime
Date
Msg-id 20010711222859.O12723@datafoundry.net
Whole thread Raw
In response to Re: vacuum and 24/7 uptime  (Mark <mark@ldssingles.com>)
List pgsql-general
On Wed, Jul 11, 2001 at 03:55:46PM -0600, Mark wrote:
:
: We increased shared memory in the linux kernel, which decreased the vacuumdb
: time from 40 minutes to 14 minutes on a 450 mhz processor.  We calculate that
: on our dual 1ghz box with ghz ethernet san connection this will go down to
: under 5 minutes.  This is acceptable to us.  Sorry about the unnecessary post.

The other thing you can do is run vacuum more frequently.  The less it
has to do, the less time it takes (it appears).  We run vacuum on an
active system every half-hour with little to no delay (the tables
aren't very big, though, like 10000 or 100000 rows, depending on the
table).  We also recreate the indices on the most frequent table every
12 hours, since vacuum doesn't clean up after indices (again, it
appears).

Of course, we're still looking forward to 7.2 where some of this will
hopefully be unnecessary.

* Philip Molter
* DataFoundry.net
* http://www.datafoundry.net/
* philip@datafoundry.net

pgsql-general by date:

Previous
From: John Clark Naldoza y Lopez
Date:
Subject: Re: A small question about Red Hat
Next
From: Tatsuo Ishii
Date:
Subject: Re: View prevents index