Date: Thu, 12 Aug 2021 18:20:06 -0500
From: Justin Pryzby <pryzby@telsasoft.com>
To: Matt Dupree <matt.dupree@heap.io>
Cc: pgsql-performance@lists.postgresql.org
Subject: Re: Postgres using the wrong index index
Message-ID: <20210812232006.GN10479@telsasoft.com>
References: 
 <CAMOk8kqSPr2sTihQ56Va07TeFRA8E1GsEUfwe4t4LSmj5pjd4A@mail.gmail.com>
 <20210811123856.GE10479@telsasoft.com>
 <CAMOk8kq5Hm80EHTtBR1xD0T0aCDmRoE3nxaWtkNSZqS04CLLfg@mail.gmail.com>
 <20210812034510.GG10479@telsasoft.com>
 <CAMOk8krqeZieBDgqRPV6HK8nXAra4D29Cc9wUpdyEa=OXH7oOw@mail.gmail.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: 
 <CAMOk8krqeZieBDgqRPV6HK8nXAra4D29Cc9wUpdyEa=OXH7oOw@mail.gmail.com>
User-Agent: Mutt/1.9.4 (2018-02-28)
Archived-At: 
 <https://www.postgresql.org/message-id/20210812232006.GN10479%40telsasoft.com>
Precedence: bulk

On Thu, Aug 12, 2021 at 09:38:45AM -0400, Matt Dupree wrote:
> > The rowcount estimate for the time column is bad for all these plans - do you
> > know why ?  You're using inheritence - have you analyzed the parent tables recently ?
> 
> Yes. I used ANALYZE before posting, as it's one of the "things to try"
> listed in the slow queries wiki. I even ran the queries immediately after
> analyzing. No difference. Can you say more about why the bad row estimate
> would cause Postgres to use the bigger index? I would expect Postgres to
> use the smaller index if it's over-estimating how many rows will be
> returned.

The overestimate is in the table's "time" column (not index) and applies to all
the plans.  Is either half of the AND estimated correctly?  If you do a query
with only ">=", and a query with only "<=", do either of them give an accurate
rowcount estimate ?

|Index Scan using other_events_1004175222_pim_evdef_67951aef14bc_idx on public.other_events_1004175222 (cost=0.28..1,648,877.92 rows=1,858,891 width=32) (actual time=1.008..15.245 rows=23 loops=1) 
|Index Cond: ((other_events_1004175222."time" >= '1624777200000'::bigint) AND (other_events_1004175222."time" <= '1627369200000'::bigint))

It seems like postgres expects the scan to return a large number of matching
rows, so tries to use the more selective index which includes the "type"
column.  But "type" is not very selective either (it has only 4 distinct
values), and "time" is not the first column, so it reads a large fraction of
the table, slowly.

Could you check pg_stat_all_tables and be sure the last_analyzed is recent for
both parent and child tables ?

Could you send the histogram bounds for "time" ?
SELECT tablename, attname, inherited, array_length(histogram_bounds,1), (histogram_bounds::text::text[])[1], (histogram_bounds::text::text[])[array_length(histogram_bounds,1)]
FROM pg_stats ... ;

-- 
Justin