Horton

Re: postgres_fdw: perform UPDATE/DELETE .. RETURNING on a join directly

Etsuro Fujita <fujita.etsuro@lab.ntt.co.jp>

View thread View on postgr.es
From: Etsuro Fujita <fujita.etsuro@lab.ntt.co.jp>
To: Andres Freund <andres@anarazel.de>
Cc: Tom Lane <tgl@sss.pgh.pa.us>, Joe Conway <mail@joeconway.com>, Ashutosh Bapat <ashutosh.bapat@enterprisedb.com>, Robert Haas <robertmhaas@gmail.com>, Pg Hackers <pgsql-hackers@postgresql.org>, Rushabh Lathia <rushabh.lathia@gmail.com>, Michael Paquier <michael.paquier@gmail.com>
Date: 2018-04-09T10:15:28Z
Lists: pgsql-hackers 
(2018/04/07 4:17), Andres Freund wrote:
> On 2018-03-05 17:07:10 -0500, Tom Lane wrote:
>> 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.
>
> I wonder if temporarily changing postgres_fdw's test to specify an extra
> config that installs auto_explain in full aggressiveness (i.e. including
> costs etc) and enables debug3 logging could help narrow this down?

+1 because we cannot deny the possibility that the plan instability is 
caused by such an unexpected behavior of postgres_fdw.

Best regards,
Etsuro Fujita

Commits

  1. Temporarily instrument postgres_fdw test to look for statistics changes.

  2. postgres_fdw: Fourth attempt to stabilize regression tests.

  3. postgres_fdw: Third attempt to stabilize regression tests.

  4. postgres_fdw: Attmempt to stabilize regression tests.

  5. postgres_fdw: Push down UPDATE/DELETE joins to remote servers.