Re: Suspected Postgres Datacorruption - Mailing list pgsql-performance

From Greg Smith
Subject Re: Suspected Postgres Datacorruption
Date
Msg-id 4E40B3D1.70707@2ndquadrant.com
Whole thread Raw
In response to Suspected Postgres Datacorruption  (Sumeet Jauhar <sumeet.jauhar@gmail.com>)
List pgsql-performance
Sumeet Jauhar wrote:
>
> Our application is running on Postgres 7.4.X . I agree that this is a
> very old version of Postgres and we should have upgraded .
>

It's important to know the .X here.  The latest 7.4 is 7.4.30:
http://www.postgresql.org/docs/7.4/static/release.html

If you're running a 7.4 much lower than .30, you almost certainly have a
version with corruption bugs related to indexes.  There's a bunch of
them mentioned in the release notes of many 7.4 versions listed there.

> I ideally want to push to a higher version of Postgres . If I can
> prove that there will be significant performance benefits and that
> crashes won’t occur then I will be able to present a strong case .
>

Go visit http://suckit.blog.hu/2009/09/29/postgresql_history for minute.

8.0 is faster than the 7.4 you're running, and that's showing the speed
increase from there.  Your application might easily run 10X as fast on a
newer PostgreSQL version.

Now, on top of all this, it sounds like you might have a problem with
your drives/controller not doing writes reliably.  See
http://wiki.postgresql.org/wiki/Reliable_Writes for more information.
If that's the situation, the version of PostgreSQL you use won't matter
too much--the database will still be unreliable if the hardware is
configured to do the wrong thing.

--
Greg Smith   2ndQuadrant US    greg@2ndQuadrant.com   Baltimore, MD
PostgreSQL Training, Services, and 24x7 Support  www.2ndQuadrant.us


pgsql-performance by date:

Previous
From: Greg Smith
Date:
Subject: Re: Postgres 8.4 memory related parameters
Next
From: "mark"
Date:
Subject: Re: benchmark woes and XFS options