Re: Preventing index scans for non-recoverable index AMs

From: Simon Riggs <simon(at)2ndQuadrant(dot)com>
To: Jaime Casanova <jcasanov(at)systemguards(dot)com(dot)ec>
Cc: pgsql-hackers <pgsql-hackers(at)postgresql(dot)org>
Subject: Re: Preventing index scans for non-recoverable index AMs
Date: 2008-12-17 22:54:28
Message-ID: 1229554468.4793.136.camel@ebony.2ndQuadrant
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-hackers


On Wed, 2008-12-17 at 17:42 -0500, Jaime Casanova wrote:
> On Wed, Dec 17, 2008 at 4:26 PM, Simon Riggs <simon(at)2ndquadrant(dot)com> wrote:
> > Hot Standby won't work with hash indexes because they are
> > non-recoverable.
> >
> > We have a number of ways of dealing with this:
> >
>
> i don't see a reason for inventing the wheel, we don't have wal for
> hash indexes because makes those more slow without any benefit at
> all... now there will be one...

Well, we're running short of time for 8.4 to put it mildly, so option
(1) is not on my radar. Even if somebody wrote WAL support for hash
indexes right now, I would be much happier with my other two suggestions
from a robustness perspective. We don't yet have a mechanism for an
index AM to say "damn, this index is screwed up, don't use it". So a
rushed implementation of WAL support would be counterproductive, ISTM.

So it's either (2), (3) or another option.

--
Simon Riggs www.2ndQuadrant.com
PostgreSQL Training, Services and Support

In response to

Responses

Browse pgsql-hackers by date

  From Date Subject
Next Message Simon Riggs 2008-12-17 22:58:11 Re: Preventing index scans for non-recoverable index AMs
Previous Message Tom Lane 2008-12-17 22:51:35 Re: Summary: changes needed in function defaults behavior