At 09:18 AM 21/11/2002 +0100, Roel Rozendaal - IC&S wrote:
There hasn't been any ANALYZE statements
at all (we migrated this db from another one and encountered the
problem afterwards); but quite frankly i would still regard it as a bug
if i need to run an ANALYZE statement first for such a simple query ;-)
Sometimes I must admit I find the optimizer/planner quite temperamental.
But usually if you work within it's rules/limitations, it does a pretty
reasonable job. I have sent details of the odd behaviour to the PG mailing
lists, and will let you know what response I get (they are usually very
responsive).
In the mean time, I can guess that one of the suggestions will be to do an
ANALYZE. This is a cheap command that does not lock anything, so it is
worth doing. Especially if, as seems to be the case, you have a test machine.
The other thing you need to do, if you are not doing it already, is a
regular vacuum of the entire database, and probably a *very* regular vacuum
of frequently UPDATED tables (eg. we vacuum the 'users' table every 5 minutes).
----------------------------------------------------------------
Philip Warner | __---_____
Albatross Consulting Pty. Ltd. |----/ - \
(A.B.N. 75 008 659 498) | /(@) ______---_
Tel: (+61) 0500 83 82 81 | _________ \
Fax: (+61) 03 5330 3172 | ___________ |
Http://www.rhyme.com.au | / \|
| --________--
PGP key available upon request, | /
and from pgp5.ai.mit.edu:11371 |/
