Dave Pirotte <dpirotte(at)mediamatters(dot)org> writes:
> We have two servers running pgsql -- an older server running 8.2.3,
> and a newer (far superior) one running 8.2.4. One of our reporting
> queries is running painfully slowly on 8.2.4, but it executes in a
> reasonable timeframe on 8.2.3.
Are you sure you've analyzed all these tables in the 8.2.4 database?
Some of the rowcount estimates seem a bit far off.
I looked through the CVS logs and didn't find any planner changes
between 8.2.3 and 8.2.4 that seem likely to affect your query, so
I'm thinking it must be a statistical discrepancy.
regards, tom lane