From: Mario Weilguni <mweilguni@sime.com>
To: Ow Mun Heng <Ow.Mun.Heng@wdc.com>
Cc: Pgsql performance <pgsql-performance@postgresql.org>
Subject: Re: Altering a column type - Most efficient way
Date: Thu, 10 Jul 2008 10:36:10 +0200
Message-ID: <4875C9FA.7020101@sime.com> (raw)
In-Reply-To: <1215676380.15319.2.camel@neuromancer.home.net>
References: <1215676380.15319.2.camel@neuromancer.home.net>
Ow Mun Heng schrieb:
> Is there any quick hacks to do this quickly? There's around
> 20-30million
> rows of data.
>
> I want to change a column type from varchar(4) to varchar()
>
> table size is ~10-15GB (and another 10-15G for indexes)
>
> What would be the preferrred way of doing it? SHould I be dropping the
> indexes 1st to make things faster? Would it matter?
>
> The punch line is that since the databases are connected via slony, this
> makes it even harder to pull it off. My last try got the DDL change
> completed in like 3 hours (smallest table of the bunch) and it hung
> everything
>
Before Postgresql supported "alter table ... type ... " conversions, I
did it a few times when I detected later that my varchar() fields were
too short, and it worked perfectly.
Example:
{OLDLEN} = 4
{NEWLEN} = 60
update pg_attribute
set atttypmod={NEWLEN}+4
where attname='the-name-of-the-column'
and attrelid=(select oid from pg_class where
relname='the-name-of-the-table')
and atttypmod={OLDLEN}+4;
This worked very well when you want to increase the maximum length,
don't try to reduce the maximum length this way!
Disclaimer: I do not know if slony might be have a problem with this.
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Reply to all the recipients using the --to and --cc options:
reply via email
To: pgsql-performance@postgresql.org
Cc: mweilguni@sime.com, Ow.Mun.Heng@wdc.com
Subject: Re: Altering a column type - Most efficient way
In-Reply-To: <4875C9FA.7020101@sime.com>
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
This inbox is served by DDX for PostgreSQL; see mirroring instructions
for how to clone and mirror all data and code used for this inbox