Re: Index is not used for "IN (non-correlated subquery)"

George <pinkisntwell(at)gmail(dot)com> writes:
> On Thu, Dec 1, 2016 at 6:58 PM, Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> wrote:
>> What
>> do you get for
>> select * from pg_stats where tablename = 'wg3ppbm_userpartner';
>> and likewise for wg3ppbm_partner?

> It is a wide table. Do you want me to dump csv here?

Shouldn't be *that* wide, with only one row in the underlying table ;-)
Maybe psql \x format would be suitable.

> In the meantime, with the help of the folks at #postgresql I was able
> to wisen up the query planner by using either one of the following two
> settings:
> SET enable_seqscan = false
> SET cpu_tuple_cost = 0.1

Well, that proves it is considering the indexscan option. But there is
something funny going on, if you have stats for these tables and yet
you're getting a default rowcount estimate.

regards, tom lane