From: | Richard Huxton <dev(at)archonet(dot)com> |
---|---|
To: | fred(at)digicamp(dot)com |
Cc: | <fred(at)digicamp(dot)com>, <pgsql-performance(at)postgresql(dot)org> |
Subject: | Re: Query optimization |
Date: | 2002-12-08 13:48:44 |
Message-ID: | 200212081348.44370.dev@archonet.com |
Views: | Raw Message | Whole Thread | Download mbox | Resend email |
Thread: | |
Lists: | pgsql-performance |
On Saturday 07 Dec 2002 8:10 pm, Fred Moyer wrote:
> Ikes, they are the same, a cut and paste error. Sorry about that. No
> joins involved, one table with 1 million records, about 255 rows, only
> about 10% of the rows contain data in this particular instance.
>
> object is indexed on active, registrant, and name as well as UPPER(name).
> Postgres version is 7.2.3
I think Jochem's got it with "enable_seqscan" - you've disabled scans so the
planner is checking one million index entries - bad idea. Try Jochem's
suggestion of re-enabling seqscan and see if that helps things along.
> db=# select count(*) from count;
> count
> ---------
> 1032953
> >> time=204790.82..204790.84 rows=10 loops=1)
> >> -> Sort (cost=nan..nan rows=1032953 width=2017) (actual
> >> time=204790.81..204790.82 rows=11 loops=1)
> >> -> Index Scan using registrant__object__idx on object
> >> (cost=0.00..81733.63 rows=1032953 width=2017) (actual
> >> time=0.14..94509.14 rows=1032946 loops=1)
> >> Total runtime: 205125.75 msec
--
Richard Huxton
From | Date | Subject | |
---|---|---|---|
Next Message | Joe Conway | 2002-12-08 19:31:54 | Re: Speeding up aggregates |
Previous Message | Jochem van Dieten | 2002-12-07 20:41:57 | Re: Query optimization |