DomainKey-Signature: a=rsa-sha1; c=nofws; d=gmail.com; s=gamma;
	h=message-id:date:from:to:subject:cc:in-reply-to:mime-version
	:content-type:content-transfer-encoding:content-disposition
	:references;
	b=EKLaRrwcgBn8U/sImloc+aMMrARfnaiODscsMgPQ+vGazt1OJ3Qr+KM1RekjwNifGg
	niJ58rN5eOqeV5agYxsDx4Vf4JJaZYIyk2lrwILow7RcO8mZO4Konkr1Q/r4yWep0Pap
	z2V2StXn3rn5+V94O5f3vYpVdH7LVUAjpzkqk=
Message-ID: <84f0acdb0808071655s79d0644o21df6b397b3616c0@mail.gmail.com>
Date: Thu, 7 Aug 2008 18:55:42 -0500
From: "Joshua Shanks" <jjshanks@gmail.com>
To: "Tom Lane" <tgl@sss.pgh.pa.us>
Subject: Re: query planner not using the correct index
Cc: "Craig Ringer" <craig@postnewspapers.com.au>,
	pgsql-performance@postgresql.org
In-Reply-To: <29507.1218152891@sss.pgh.pa.us>
MIME-Version: 1.0
Content-Type: text/plain; charset=ISO-8859-1
Content-Transfer-Encoding: 7bit
Content-Disposition: inline
References: <84f0acdb0808061435q5ebb607cjc311f1eb4da841a@mail.gmail.com>
	<489A623A.4090400@postnewspapers.com.au>
	<84f0acdb0808062014p29cf7482ied332225a0a92c63@mail.gmail.com>
	<489A6AF5.50908@postnewspapers.com.au>
	<84f0acdb0808070811l43a3da9eq240d272f8ddd105c@mail.gmail.com>
	<27696.1218143560@sss.pgh.pa.us>
	<84f0acdb0808071501n15b0182ga7eca7b1be37d705@mail.gmail.com>
	<29507.1218152891@sss.pgh.pa.us>

Yeah with default_statistics_target at 500 most_common_vals had 4
values with the fourth having a frequency of 1.5% and distinct have
250+ in it.

How do I increase the stats target for just one column?

On Thu, Aug 7, 2008 at 6:48 PM, Tom Lane <tgl@sss.pgh.pa.us> wrote:
> "Joshua Shanks" <jjshanks@gmail.com> writes:
>> SELECT null_frac, n_distinct, most_common_vals, most_common_freqs FROM
>> pg_stats WHERE tablename = 'bars' AND attname='bars_id';
>>  null_frac | n_distinct |   most_common_vals   |     most_common_freqs
>> -----------+------------+----------------------+---------------------------
>>          0 |         14 | {145823,47063,24895} | {0.484667,0.257333,0.242}
>
>> Those 3 values in reality and in the stats account for 98% of the
>> rows. actual distinct values are around 350
>
> So you need to increase the stats target for this column.  With those
> numbers the planner is going to assume that any value that's not one
> of the big three appears about (1 - (0.484667+0.257333+0.242)) / 11
> of the time, or several hundred times in 300K rows.  If n_distinct were
> up around 350 it would be estimating just a dozen or so occurrences,
> which should push the join plan into the shape you want.  It's likely
> that it won't bother to include any more entries in most_common_vals
> no matter how much you raise the target; but a larger sample should
> definitely give it a better clue about n_distinct.
>
>                        regards, tom lane
>