Message-ID: <52138785.8010607@fuzzy.cz>
Date: Tue, 20 Aug 2013 17:13:09 +0200
From: Tomas Vondra <tv@fuzzy.cz>
User-Agent: Mozilla/5.0 (X11; Linux x86_64;
 rv:17.0) Gecko/20130801 Thunderbird/17.0.8
MIME-Version: 1.0
To: pgsql-performance@postgresql.org
Subject: Re: queries with DISTINCT / GROUP BY giving different plans
References: <3a56c4d2c0b2daa11a625c2983823901.squirrel@sq.gransy.com>
 <11586.1376505314@sss.pgh.pa.us> <520E7F33.7010305@fuzzy.cz>
In-Reply-To: <520E7F33.7010305@fuzzy.cz>
Content-Type: text/plain; charset=ISO-8859-2
Content-Transfer-Encoding: 7bit
Precedence: bulk
Sender: pgsql-performance-owner@postgresql.org

On 16.8.2013 21:36, Tomas Vondra wrote:
>
> Now, the difference between 24 and 56 is caused by hash_agg_entry_size.
> It's called with numAggs=0 but returns 32. I'm wondering if it should
> return 0 in such cases, i.e. something like this:
> 
>     Size
>     hash_agg_entry_size(int numAggs)
>     {
>         Size        entrysize;
> 
>         if (numAggs == 0)
>             return 0;
> 
>         /* This must match build_hash_table */
>         entrysize = sizeof(AggHashEntryData) +
>             (numAggs - 1) * sizeof(AggStatePerGroupData);
>         entrysize = MAXALIGN(entrysize);
>         /* Account for hashtable overhead (assuming fill factor = 1) */
>         entrysize += 3 * sizeof(void *);
>         return entrysize;
>     }
> 
> I've tested that after this both queries use HashAggregate (which is the
> right choice), but I haven't done any extensive checking so maybe I'm
> missing something.

So, is this a sufficient / correct explanation? Any comments about the
fix I suggested? Or should I try to get a permission to provide the data
so that you can reproduce the issue on your own? That might take a few
days to get through.

Tomas


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