Hi Tom,
Sorry, I didn't ask the right question. I meant to ask "Why does it estimate a smaller cost for the seqscan?"
With some further staring I was able to find the bad estimate and fix it by increasing the relevant statistics target.
Thanks,
Brian
----- Original Message ----
From: Tom Lane <tgl@sss.pgh.pa.us>
To: Brian Herlihy <btherl@yahoo.com.au>
Cc: Postgresql Performance <pgsql-performance@postgresql.org>
Sent: Wednesday, 14 February, 2007 4:53:54 PM
Subject: Re: [PERFORM] An unwanted seqscan
Brian Herlihy <btherl@yahoo.com.au> writes:
> I am having trouble understanding why a seqscan is chosen for this query.
As far as anyone can see from this output, the planner's decisions are
correct: it prefers the plans with the smaller estimated cost. If you
want us to take an interest, provide some more context --- EXPLAIN
ANALYZE output for starters.
regards, tom lane
---------------------------(end of broadcast)---------------------------
TIP 3: Have you checked our extensive FAQ?
http://www.postgresql.org/docs/faq