On Wed, 9 Mar 2005 11:08 pm, Hugo Ferreira wrote:
> For example... I run the query, it takes 122seconds. Then I delete the
> target tables, vacuum the database, re-run it again: 9s. But if I run
> vacuum several times, and then run, it takes again 122seconds. If I
> stop this 122seconds query, say, at second 3 and then run it again, it
> will only take 9s. It simply doesn't make sense. Also, explain analyse
> will give me diferent plans each time I run it... Unfortunately, this
> is rendering PostgreSQL unusable for our goals. Any ideas?
>
The explain analyze is still be best information if you want assistance with
what postgresql is doing, and how to stop it. If you could attach
explain analyzes for both the fast (9s), and slow (122s) runs, that would
help people get an idea of how the query is running. At the moment
we don't know how postgresql is actually executing the query.
Regards
Russell Smith.