Re: fsm and vacuum

Somehow this moved to the pgadmin list. It was intended for pgsql-admin. My apologies.
This is a dba task, I'd never expect pgadmin would do this.

From: Michael Shapiro [mailto:mshapiro51(at)gmail(dot)com]
Sent: Friday, December 03, 2010 9:26 AM
To: Guillaume Lelarge
Cc: Little, Douglas; PgAdmin Support
Subject: Re: [pgadmin-support] fsm and vacuum

I understand, but in this case, since the option is offered next to the safe one, most people won't know it isn't safe.
I certainly didn't until I read this posting. I know generally what vacuuming does, but I had no idea that postgres offered a potentially damaging option. Also, PgAdmin sometimes tells me that a table needs vacuuming, so it is already "advising" people in that area ...
On Fri, Dec 3, 2010 at 9:19 AM, Guillaume Lelarge <guillaume(at)lelarge(dot)info<mailto:guillaume(at)lelarge(dot)info>> wrote:
Le 03/12/2010 15:17, Michael Shapiro a écrit :
> The document http://wiki.postgresql.org/wiki/VACUUM_FULL says:
>
> VACUUM FULL, unlike VACUUM, tuples data that has not been deleted, moving it
> into spaces earlier in the file that have been freed. Once it's created a
> free space at the end of the file, it truncates the file so that the OS
> knows that space is free and may be reused for other things. Moving in-use
> data around this way has some major downsides and side-effects, especially
> the way VACUUM FULL does it. There are better ways to free space if you need
> to and better ways to optimize tables (see below) so *you should essentially
> never use VACUUM FULL*.
>
>
> PgAdmin does not give the user a comparable warning when it goes to execute
> a VACCUM FULL. Given the potential problems with the FULL option, would it
> make sense for PgAdmin to issue a warning to this effect?
>

I'm not sure this is the role of pgAdmin to warn people they are doing
potentially stupid things.

--
Guillaume
http://www.postgresql.fr
http://dalibo.com