Re: 7.4 dramatically slower than 7.3?

Nabil Sayegh <postgresql(at)e-trolley(dot)de> writes:
> Stephan Szabo wrote:
>> One other possibility is that the planning is taking a really long time.
>> How long do explain (non-analyze) on the two systems take? And how many
>> joins are there precisely?

> Yeah, that seems to be it.
> It takes ~
> (real) 3.776s on 7.4
> (real) 0.082s on 7.3
> measured with bash/time but shouldn't make a difference

> So the plan is actually better but takes much longer to ... plan, right?

Right. In 7.3 your use of JOIN syntax limited the planner's search for
good plans; in 7.4 it doesn't do so (by default anyway). You may care
to read http://www.postgresql.org/docs/7.4/static/explicit-joins.html
and experiment with altering JOIN_COLLAPSE_LIMIT (and/or reordering the
joins in your query).

regards, tom lane