Thanks for your reply.
On Mon, Feb 4, 2013 at 3:48 PM, Tom Lane <tgl@xxxxxxxxxxxxx> wrote:
You're worrying about the wrong thing entirely. The right thing to be
worrying about is why are some of those row estimates off by four orders
of magnitude, and what you can do to fix that. The planner will never
deliver sane plans when its estimates are so far off from reality.
You mean the rows from explain is bigger then the rows real returned?
It caused by a sub query with a filter like ( a and b ) or ( c and d ), I've tried to create a temp table for it, the explained rows and the real rows will similarly.
>From my perspective, you snipped all the interesting parts of this
output, because the misestimation is evidently happening somewhere down
in there.
regards, tom lane
