Re: get_actual_variable_range vs idx_scan/idx_tup_fetch

From: Bruce Momjian <bruce(at)momjian(dot)us>
To: Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us>
Cc: Marko Tiikkaja <marko(at)joh(dot)to>, PostgreSQL-development <pgsql-hackers(at)postgresql(dot)org>
Subject: Re: get_actual_variable_range vs idx_scan/idx_tup_fetch
Date: 2014-10-18 13:58:10
Message-ID: 20141018135810.GE25696@momjian.us
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-hackers

On Fri, Oct 17, 2014 at 11:03:04PM -0400, Tom Lane wrote:
> Bruce Momjian <bruce(at)momjian(dot)us> writes:
> > On Fri, Oct 17, 2014 at 06:15:37PM -0400, Tom Lane wrote:
> >> Those stats were perfectly valid: what the planner is looking for is
> >> accurate minimum and maximum values for the index's leading column, and
> >> that's what it got. You're correct that a narrower index could have given
> >> the same results with a smaller disk footprint, but the planner got the
> >> results it needed from the index you provided for it to work with.
>
> > Uh, why is the optimizer looking at the index on a,b,c and not just the
> > stats on column a, for example? I am missing something here.
>
> Because it needs up-to-date min/max values in order to avoid being
> seriously misled about selectivities of values near the endpoints.
> See commit 40608e7f949fb7e4025c0ddd5be01939adc79eec.

Oh, I had forgotten we did that. It is confusing that there is no way
via EXPLAIN to see the access, making the method of consulting pg_stat_*
and using EXPLAIN unreliable. Should we document this somewhere?

--
Bruce Momjian <bruce(at)momjian(dot)us> http://momjian.us
EnterpriseDB http://enterprisedb.com

+ Everyone has their own god. +

In response to

Responses

Browse pgsql-hackers by date

  From Date Subject
Next Message Bruce Momjian 2014-10-18 14:23:52 Re: Code bug or doc bug?
Previous Message Nicolas Barbier 2014-10-18 13:06:16 Re: Materialized views don't show up in information_schema