Subject: Re: Index bloat of 4x
From: Csaba Nagy <nagy@ecircle-ag.com>
To: "Ed L." <pgsql@bluepolka.net>
Cc: Postgres general mailing list <pgsql-general@postgresql.org>,
	Bill Moran <wmoran@collaborativefusion.com>, Tom Lane <tgl@sss.pgh.pa.us>,
	Alvaro Herrera <alvherre@commandprompt.com>
In-Reply-To: <200701181842.00696.pgsql@bluepolka.net>
References: <20070117093223.bcdb302e.wmoran@collaborativefusion.com>
	<19674.1169104105@sss.pgh.pa.us>
	<20070118080715.479b9315.wmoran@collaborativefusion.com>
	<200701181842.00696.pgsql@bluepolka.net>
Content-Type: text/plain
Message-Id: <1169197877.11526.10.camel@coppola.muc.ecircle.de>
Mime-Version: 1.0
Date: Fri, 19 Jan 2007 10:11:17 +0100
Content-Transfer-Encoding: 7bit

[snip]
> I afraid I don't see how any of the answers I saw discussed fit a 
> 24x7 operation.  Reindex, drop index, vacuum full, ... they all 
> block production queries of one sort or another for significant 
> periods of time (minutes) on large (multi/tens of GB) tables, 
> and thus are infeasible for true 24x7 operations.[snip]

This is not completely true, as of 8.2 there is an online index build,
and if that could be used in a background thread to rebuild the index
and replace the bloated one once it's finished, that would be a
non-blocking operation which could be done in 24x7 situations.

There are some issues with using the online index build for replacing an
existing index, but if those could be solved it would be a viable
solution I think...

Cheers,
Csaba.