Tom Lane <tgl@sss.pgh.pa.us> wrote:
> I'm not sure how much it would help to increase the statistics
> targets, but that would be worth trying.
I notice that the scan rowcount estimates are very accurate, there's
that one hash join result that's way off, though.
What's up with the sort of _accrged7200 (in the slower plan) taking
in 3.5 million rows and putting out 1 row? There's something there
I'm not understanding.
-Kevin