From: Tom Lane <tgl@sss.pgh.pa.us>
To: a.schnabl@synedra.com
cc: pgsql-bugs@lists.postgresql.org
Subject: Re: BUG #16993: Query Planner does not use index for EXISTS-query on
 a large table with all equal foreign key values
In-reply-to: <16993-1298e8dc9c41ab97@postgresql.org>
References: <16993-1298e8dc9c41ab97@postgresql.org>
Comments: In-reply-to PG Bug reporting form <noreply@postgresql.org>
	message dated "Wed, 05 May 2021 13:24:05 -0000"
MIME-Version: 1.0
Content-Type: text/plain; charset="us-ascii"
Content-ID: <4016382.1620225070.1@sss.pgh.pa.us>
Date: Wed, 05 May 2021 10:31:10 -0400
Message-ID: <4016383.1620225070@sss.pgh.pa.us>
Archived-At: 
 <https://www.postgresql.org/message-id/4016383.1620225070%40sss.pgh.pa.us>
Precedence: bulk

PG Bug reporting form <noreply@postgresql.org> writes:
> ... EXPLAIN ANALYZE shows the following query plan:

> Merge Join  (cost=179053.41..179054.24 rows=1 width=8) (actual
> time=2140.820..2140.822 rows=1 loops=1)

> [ but with ]  SET enable_seqscan = false:

> Nested Loop Semi Join  (cost=0.57..12.85 rows=1 width=8) (actual
> time=0.877..0.920 rows=1 loops=1)

Yeah, that seems *extremely* broken.  The planner correctly estimates
the second plan as much cheaper, so why did it fail to choose that one
without help?  Will look into this.

			regards, tom lane