From: Jerry Sievers <gsievers19@comcast.net>
To: Rajendra prasad <rajendra.dn@gmail.com>
Cc: pgsql-admin@postgresql.org
Subject: Re: Index size growing
References: <AANLkTik6kMGP=k7x_Eyk0ydwVVv_zKLYy6P45UN0BDoX@mail.gmail.com>
Date: Fri, 01 Apr 2011 07:28:31 -0400
In-Reply-To: <AANLkTik6kMGP=k7x_Eyk0ydwVVv_zKLYy6P45UN0BDoX@mail.gmail.com>
	(Rajendra prasad's message of "Fri, 1 Apr 2011 07:31:40 +0530")
Message-ID: <871v1mcisg.fsf@comcast.net>
User-Agent: Gnus/5.13 (Gnus v5.13) Emacs/23.1 (gnu/linux)
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii

Rajendra prasad <rajendra.dn@gmail.com> writes:

> Hi,
>
> I have 50 GB data base folder for postgres. Out of 50 GB 20 GB has
> been occupied by 8 indexes of a single table by growing upto 2 GB
> each. I googled and got the info that reindex will help brining back
> the actual size of the index. But the disk has only 4 GB left . Due to
> this, when i run reindex it is getting almost full. I don't know
> exactly why it is growing like that during reindex. Due to this i am
> not able to complete the reindex and see if i get back the disk
> spacce. Please suggest me a good solution and the way that i am doing
> is logical or not.

Did you try reindexing each index individually?

The old index files will be dropped in that case upon completion of each
command if you are running autocommit.  Less transient disk space demand
by that strategy.

Even more disk friendly is just to drop/create the indexes.

Since you can't reindex without blocking anyway, I assume you are clear
of concurrency issues to prevent using this advice (apps down and/or
quiescent).

-- 
Jerry Sievers
Postgres DBA/Development Consulting
e: gsievers19@comcast.net
p: 305.321.1144