Skip site navigation (1) Skip section navigation (2)

Best practise for upgrade of 24GB+ database

From: francis picabia <fpicabia(at)gmail(dot)com>
To: pgsql-admin(at)postgresql(dot)org
Subject: Best practise for upgrade of 24GB+ database
Date: 2012-01-20 18:12:13
Message-ID: (view raw, whole thread or download thread mbox)
Lists: pgsql-admin
In an academic setting, we have a couple of larger than typical
Postgres databases.
One for moodle is now 15GB and another for a research project is
currently 24 GB.

I notice while upgrading Postgresql in Debian from 8.3 to 8.4, the downtime
on the 24 GB research database is extensive while using pg_upgradecluster

It has now been 26 hours of downtime for the database, and about 18GB of
the 24GB is recovered into the 8.4 destination so far.

I read some of the tips on the Postgresql wiki on performance tweaks
( )
and had implemented some improvements such as shared_buffers
in the 8.3 instance prior to the upgrade.  I thought if I was doing this
again, I would have found the source postgresql.conf used by
the pg_upgradecluster script for 8.4, and tuned it prior to the run.

How do others manage larger database upgrades while minimizing
downtime?  Do you avoid pg_upgradecluster and simply do a pg_restore
from a dump made prior to the upgrade?  Do you run a replication
and then resync it after the upgrade is complete?  Googling for info
on this I've only found remarks about it taking longer than you'd expect.


pgsql-admin by date

Next:From: Nicholson, Brad (Toronto, ON, CA)Date: 2012-01-20 18:45:07
Subject: Re: Best practise for upgrade of 24GB+ database
Previous:From: Kevin GrittnerDate: 2012-01-20 15:38:34
Subject: Re: Unable to Connect to server

Privacy Policy | About PostgreSQL
Copyright © 1996-2017 The PostgreSQL Global Development Group