"Good, Thomas" <tgood@svcmcny.org> writes:
> I have a question about a query that starts out fine and over time slows to a halt - but only on a webhosted site.
Locallyit does fine.
> The query is a singleton select (no joins), hitting a table with about
> 5,000 records in it. Over time the query slows to a crawl and I have to
> dump and rebuild/repopulate to restore efficiency. Vacuum does nothing
> but a dump and reload fixes the problem and the query runs lightning
> fast again -- for a period of a week or so.
If a dump and reload fixes it then you have table-bloat or index-bloat
problems. A simple VACUUM won't fix that once it's happened. You need
to VACUUM more often so that it doesn't get bloated in the first place.
> Locally I'm running 8.4.2, the webhost is 8.1.18
... and the reason for the difference is probably that 8.4.x has a
reasonably decent autovacuum facility, which 8.1.x does not. Suggest
bugging your webhost provider to provide a less obsolete version of
Postgres.
regards, tom lane