From: Tom Lane <tgl@sss.pgh.pa.us>
To: Joe Conway <mail@joeconway.com>
cc: Ashutosh Bapat <ashutosh.bapat@enterprisedb.com>,
        Etsuro Fujita <fujita.etsuro@lab.ntt.co.jp>,
        Robert Haas <robertmhaas@gmail.com>,
        Pg Hackers <pgsql-hackers@postgresql.org>,
        Rushabh Lathia <rushabh.lathia@gmail.com>,
        Michael Paquier <michael.paquier@gmail.com>
Subject: Re: postgres_fdw: perform UPDATE/DELETE .. RETURNING on a join
 directly
In-reply-to: <857d2c90-492a-028e-37b9-66d86d8cd5dc@joeconway.com>
References: <5A438A45.2000909@lab.ntt.co.jp> <5A7BE8E6.200@lab.ntt.co.jp>
 <28606.1518105934@sss.pgh.pa.us>
 <CA+Tgmobf+AXEwO4r0oCkvNhkaVnxPgoJaApoX+rAi0G7-N-Dyg@mail.gmail.com>
 <5A7CFDDA.8050906@lab.ntt.co.jp> <5A7D76CF.8080601@lab.ntt.co.jp>
 <CA+TgmoaAQCn0+xh3JG2avn35Lukj2NVmhxVi7hE2CwzSUQ-5ag@mail.gmail.com>
 <20134.1518201188@sss.pgh.pa.us>
 <CA+TgmoY5KNf-Y=zOdHVfa8Lmo=OMJe=dBpP-noT_1fdVK6EYPA@mail.gmail.com>
 <18653.1518297871@sss.pgh.pa.us> <5A82DCCE.3060107@lab.ntt.co.jp>
 <29147.1519769644@sss.pgh.pa.us>
 <CA+Tgmob_WGK+vKzE6FmrMmuwuv4myntfRAkEbo+tsDt9xUwCLQ@mail.gmail.com>
 <23783.1519929258@sss.pgh.pa.us> <5A9928F1.2010206@lab.ntt.co.jp>
 <CA+TgmoaNWpQ+F7_DtiJpc62PUJ6JQBAQpT1H7CBrjKmryGGgkw@mail.gmail.com>
 <CA+TgmoZOpGnmss2dpOAktHc30NS=StmLw-ChkJkd74DLo_nyEA@mail.gmail.com> <! !
 5A9CB7B9.2040008@lab.ntt.co.jp>
 <CAFjFpRdQVqCrYXN3F=jj_wftSp2gD-rPXKS-ogoQLAmrmKxwuQ@mail.gmail.com>
 <21358.1520229112@sss.pgh.pa.us> <25502.1520277552@sss.pgh.pa.us>
 <857d2c90-492a-028e-37b9-66d86d8!
 cd5dc@joeconway.com>
Comments: In-reply-to Joe Conway <mail@joeconway.com>
	message dated "Mon, 05 Mar 2018 12:11:46 -0800"
MIME-Version: 1.0
Content-Type: text/plain; charset="us-ascii"
Content-ID: <22163.1520287630.1@sss.pgh.pa.us>
Content-Transfer-Encoding: quoted-printable
Date: Mon, 05 Mar 2018 17:07:10 -0500
Message-ID: <22164.1520287630@sss.pgh.pa.us>
Precedence: bulk

Joe Conway <mail@joeconway.com> writes:
> On 03/05/2018 11:19 AM, Tom Lane wrote:
>> Joe, I wonder if you could add "log_autovacuum_min_duration =3D 0" to
>> rhinoceros' extra_config options, temporarily?  Correlating that log
>> output with the log_statement output from the test proper would let
>> us confirm or deny whether it's autovacuum.

> Done just now. Do you want me to force a run?

Both of these runs

https://buildfarm.postgresql.org/cgi-bin/show_log.pl?nm=3Drhinoceros&dt=3D=
2018-03-05%2020%3A35%3A00
https://buildfarm.postgresql.org/cgi-bin/show_log.pl?nm=3Drhinoceros&dt=3D=
2018-03-05%2021%3A45%3A02

appear to exonerate autovacuum, and the second seems to destroy the
behind-the-scenes-ANALYZE theory entirely, since there was no change
in the outputs of the extra instrumentation queries.  (In theory
there's a window for ANALYZE to have occurred in between, but that's
not very credible.)

So you can revert the rhinoceros config change if you like --- thanks
for making it so quickly!

Meanwhile, I'm back to wondering what could possibly have affected
the planner's estimates, if pg_proc and pg_statistic didn't change.
I confess bafflement ... but we've now eliminated the autovacuum-
did-it theory entirely, so it's time to start looking someplace else.
I wonder if something in the postgres_fdw remote join machinery
is not as deterministic as it should be.

			regards, tom lane