Re: Disk Utilization Increases And Time for Vacuum Increases. - Mailing list pgsql-admin

From Curt Sampson
Subject Re: Disk Utilization Increases And Time for Vacuum Increases.
Date
Msg-id Pine.NEB.4.51.0308281108310.1359@angelic-vtfw.cvpn.cynic.net
Whole thread Raw
In response to Re: Disk Utilization Increases And Time for Vacuum Increases.  (Andrew Sullivan <andrew@libertyrms.info>)
Responses Re: Disk Utilization Increases And Time for Vacuum Increases.
List pgsql-admin
On Tue, 26 Aug 2003, Andrew Sullivan wrote:

> I'm afraid that's the problem, though.  The code cannot recover freed
> btree pages which can't be recycled for others to use.  There is no
> choice but to reindex certain indexes.

While we're at it, anybody got a clue on how to reindex system tables?
Our unit testing framework (yes, we unit test all of our database code)
has to do a lot of schema loads and drops, and the indexes for the
system tables (particularly pg_attribute) tend to grow dramatically,
slowing performance. (The total size of the pg_catalog starts out around
a few megabytes, and doesn't take too long to grow to several hundred
megabytes.)

So far, the only way I've found to fix this is to do a complete dump
and reload of the database, but that's a pain since we have tables with
mutual constraints that can't be reloaded without modifying the dump
file.

cjs
--
Curt Sampson  <cjs@cynic.net>   +81 90 7737 2974   http://www.NetBSD.org
    Don't you know, in this new Dark Age, we're all light.  --XTC

pgsql-admin by date:

Previous
From: Hornyak Laszlo
Date:
Subject: Re: postgres hab modification without restarting
Next
From: Chris Miles
Date:
Subject: Re: postgresql and replication