From: Tom Lane <tgl@sss.pgh.pa.us>
To: Richard Neill <rn214@richardneill.org>
cc: "pgsql-performance@postgresql.org" <pgsql-performance@postgresql.org>
Subject: Re: Why does the query planner use two full indexes,
	when a dedicated partial index exists?
In-reply-to: <50D2A8BA.2070801@richardneill.org>
References: <50D22DE2.4090301@richardneill.org>
	<CAL_0b1s6_YZDdzt3NizwC8X+dYA37T-4aZOTzw3KJ3-iXuSjjw@mail.gmail.com>
	<50D2528E.7000800@richardneill.org>
	<CAL_0b1t5_h-VUE+sL2cfd1E78xYOq9T7P3n31iZhLT3vEa=mOA@mail.gmail.com>
	<50D25A59.3060105@richardneill.org>
	<CAMkU=1wYVjf63j+4ttZrVMmYfst7Q7ngez5e6X4s+keYbpcwdg@mail.gmail.com>
	<50D2A8BA.2070801@richardneill.org>
Comments: In-reply-to Richard Neill <rn214@richardneill.org>
	message dated "Thu, 20 Dec 2012 05:57:14 +0000"
Date: Thu, 20 Dec 2012 10:43:26 -0500
Message-ID: <11997.1356018206@sss.pgh.pa.us>
Precedence: bulk
Sender: pgsql-performance-owner@postgresql.org

Richard Neill <rn214@richardneill.org> writes:
> Also, I wonder whether it matters which order the indexes are created in?

IIRC, if the estimated costs of using two different indexes come out the
same (to within 1% or so), then the planner keeps the first-generated
path, which will result in preferring the index with smaller OID.  This
effect doesn't apply to your problem query though, since we can see from
the drop-experiments that the estimated costs are quite a bit different.

A more likely explanation if you see some effect that looks like order
dependency is that the more recently created index has accumulated less
bloat, and thus has a perfectly justifiable cost advantage.

			regards, tom lane


-- 
Sent via pgsql-performance mailing list (pgsql-performance@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-performance