Re: [HACKERS] Another index "buglet"?

The Hermit Hacker <scrappy(at)hub(dot)org> writes:
> Forbid merge:
> Hash Join (cost=9994.31 rows=2740488 width=36)
> -> Seq Scan on url (cost=3368.58 rows=37866 width=20)
> -> Hash (cost=3368.58 rows=37866 width=16)
> -> Seq Scan on url url2 (cost=3368.58 rows=37866 width=16)

> 0.900u 0.217s 0:04.19 26.4% 103+14638k 0+175io 0pf+0w

> Forbid Hash:
> Merge Join (cost=11188.76 rows=2740488 width=36)
> -> Index Scan using url_pkey on url url2 (cost=4347.30 rows=37866 width=16)
> -> Index Scan using url_referrer on url (cost=4342.30 rows=37866 width=20)

> 0.897u 0.210s 0:03.19 34.4% 106+15120k 0+179io 0pf+0w

Thanks, but I'm confused about what I'm looking at here. Are those
time outputs for the backend, or for psql?

Also, how large are these two tables, and how many rows do you actually
get from the query?

regards, tom lane