From: Tomas Vondra <tomas.vondra@2ndquadrant.com>
To: Tom Lane <tgl@sss.pgh.pa.us>
Cc: Kamil Jońca <kjonca@fastmail.com>
Cc: pgsql-bugs@lists.postgresql.org
Subject: Re: backend crash
Date: Mon, 13 Apr 2020 17:32:46 +0200
Message-ID: <20200413153246.r5amtd62cxvwlqkn@development> (raw)
In-Reply-To: <394.1586790061@sss.pgh.pa.us>
References: <87imi37spc.fsf@alfa.kjonca>
<20200413114147.a42vf3tr3auybdtr@development>
<394.1586790061@sss.pgh.pa.us>
On Mon, Apr 13, 2020 at 11:01:01AM -0400, Tom Lane wrote:
>Tomas Vondra <tomas.vondra@2ndquadrant.com> writes:
>> On Mon, Apr 13, 2020 at 09:25:35AM +0200, Kamil Jońca wrote:
>>> I have problem with (pseudo) random backend crashes.
>>> They seems to happen at one query, but not every time.
>
>> I've tried to reproduce this, but unfortunately it does not fail for me.
>> Chances are this depends on data. Can you prepare a small data sample
>> that would trigger the issue?
>
>This is evidently a parallel-query worker, so it's unlikely you'd ever
>get anywhere near this crash with a "small" data sample; for starters
>there'd have to be enough data to prompt use of a parallel plan.
>
>I doubt that the crash is very data-specific though. Maybe just fill
>the table with a lot of dummy data?
>
I did that, of course, and the plan looks like this:
QUERY PLAN
-----------------------------------------------------------------------------------------------------------------------------------------------------
Finalize Aggregate (cost=315448.35..315448.36 rows=1 width=8) (actual time=18685.398..18685.400 rows=1 loops=1)
-> Gather (cost=315447.52..315448.33 rows=8 width=8) (actual time=18683.862..18687.698 rows=8 loops=1)
Workers Planned: 8
Workers Launched: 7
-> Partial Aggregate (cost=314447.52..314447.53 rows=1 width=8) (actual time=18669.777..18669.781 rows=1 loops=8)
-> Result (cost=0.00..299447.50 rows=12500000 width=40) (actual time=0.160..17844.793 rows=250000 loops=8)
-> ProjectSet (cost=0.00..80697.50 rows=12500000 width=32) (actual time=0.145..14854.132 rows=250000 loops=8)
-> Parallel Seq Scan on log (cost=0.00..16947.50 rows=125000 width=81) (actual time=0.026..343.471 rows=125000 loops=8)
Filter: (account = 'onet'::text)
Planning Time: 0.223 ms
Execution Time: 18687.862 ms
(11 rows)
but still no crashes. I've even built from the exact 12.2 commit to make
sure the bug was not accidentally fixed by some commit since then.
regards
--
Tomas Vondra http://www.2ndQuadrant.com
PostgreSQL Development, 24x7 Support, Remote DBA, Training & Services
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Reply to all the recipients using the --to and --cc options:
reply via email
To: pgsql-bugs@postgresql.org
Cc: tomas.vondra@2ndquadrant.com, tgl@sss.pgh.pa.us, kjonca@fastmail.com, pgsql-bugs@lists.postgresql.org
Subject: Re: backend crash
In-Reply-To: <20200413153246.r5amtd62cxvwlqkn@development>
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
This inbox is served by DDX for PostgreSQL; see mirroring instructions
for how to clone and mirror all data and code used for this inbox