Re: Suboptimal plan choice problem with 8.3RC2

On Jan 22, 2008 9:52 PM, Alvaro Herrera <alvherre@commandprompt.com> wrote:
> No, autovacuum should have taken care of it.  I would be interesting in
> knowing why it didn't.

I just dropped the database on monday morning and import it again.
Nothing really fancy.

It seems that autovacuum took care of them just after the import which
is what I expected:

cityvox_prod=# select relname, last_vacuum, last_autovacuum,
last_analyze, last_autoanalyze from pg_stat_all_tables where
schemaname = 'cityvox' AND relname IN('association', 'evelieu',
'assovil', 'lieu', 'vilquartier') ORDER BY relname;  relname   | last_vacuum | last_autovacuum |         last_analyze
 |       last_autoanalyze
 

-------------+-------------+-----------------+-------------------------------+-------------------------------association
|            |                 | 2008-01-22
 
20:34:26.813283+01 | 2008-01-21 12:10:50.30652+01assovil     |             |                 | 2008-01-22
20:34:46.548442+01 | 2008-01-21 12:10:50.573546+01evelieu     |             |                 | 2008-01-22
20:34:33.193569+01 | 2008-01-21 12:11:06.237325+01lieu        |             |                 | 2008-01-22
20:34:35.936066+01 | 2008-01-21 12:11:52.085856+01vilquartier |             |                 | 2008-01-22
20:34:43.409459+01 | 2008-01-21 12:12:08.391397+01

So I wonder why the stats were so bad... I didn't update the data at
all after the initial import. Any idea?

I'll check the stats of these tables next time I drop/recreate the database.

--
Guillaume