I've never received any reply to this post; as I said, I think I have a dump that recreates the problem.
--- Ven 14/8/09, Scara Maccai <m_lists@xxxxxxxx> ha scritto:
> Da: Scara Maccai <m_lists@xxxxxxxx>
> Oggetto: Re: totally different plan when using partitions
> A: "Tom Lane" <tgl@xxxxxxxxxxxxx>
> Cc: "pgsql-general" <pgsql-general@xxxxxxxxxxxxxx>
> Data: Venerdì 14 agosto 2009, 09:43
>
>
> Query:
> set enable_mergejoin=off;set enable_hashjoin=off;
>
> explain analyze
> select nome1,
> thv3tralacc,
> dltbfpgpdch
> FROM cell_bsc_60_0610 as cell_bsc
> left outer join
> teststscell73_test as data on data.ne_id=cell_bsc.nome1
> left outer join
> teststscell13_test as data1 on data1.ne_id=cell_bsc.nome1
> and data1.time=data.time
> where
>
> data.time >=cell_bsc.starttime and
> data.time <=cell_bsc.endtime and
> data.time between '2006-10-01
> 00:00:00' and '2006-10-02 01:00:00'
> and data1.time >=cell_bsc.starttime
> and data1.time <=cell_bsc.endtime
> and
> data1.time between '2006-10-01 00:00:00' and '2006-10-02
> 01:00:00'
> and cell_bsc.nome2=2;
>
> Explain analyze on my sistem ("PostgreSQL 8.4.0 on
> sparc-sun-solaris2.10, compiled by cc: Sun C 5.9 SunOS_sparc
> Patch 124867-02 2007/11/27, 64-bit"):
>
> http://explain-analyze.info/query_plans/3822-query-plan-2528
>
> interesting bit (if I got it right):
>
> -> Nested Loop (cost=0.0..3139.65 rows=6531 width=36)
> (actual time=0.141..13.459 rows=3692 loops=1)
> Join Filter: ((data1.time >=
> cell_bsc.starttime) AND (data1.time <= cell_bsc.endtime)
> AND (cell_bsc.nome1 = data1.ne_id))
> -> Seq Scan on
> cell_bsc_60_0610 cell_bsc (cost=0.0..99.39 rows=285
> width=20) (actual time=0.033..1.740 rows=285 loops=1)
> Filter:
> (nome2 = 2)
> -> Append
> (cost=0.0..10.6 rows=4 width=16) (actual time=0.012..0.027
> rows=13 loops=285)
>
>
> 285*4 should give 1140 max, not 6531: is this different
> enough?
>
> If it is, I have a 585K dump file that should recreate the
> problem.
>
>
>
>
>
--
Sent via pgsql-general mailing list (pgsql-general@xxxxxxxxxxxxxx)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-general
