Re: RT3.4 query needed a lot more tuning with 9.2 than it did with 8.1

Robert Haas <robertmhaas@gmail.com> writes:
> The planner is estimating this the outer side of this nested loop will
> produce 33 rows and that the inner side will produce 1.  One would
> assume that the row estimate for the join product couldn't be more
> than 33 * 1 = 33 rows, but the planner is estimating 62335 rows, which
> seems like nonsense.

You know, of course, that the join size estimate isn't arrived at that
way.  Still, this point does make it seem more like a planner bug and
less like bad input stats.  It would be nice to see a self-contained
example ...

            regards, tom lane