Horton

Re: BUG #17485: Records missing from Primary Key index when doing REINDEX INDEX CONCURRENTLY

Alvaro Herrera <alvherre@alvh.no-ip.org>

View thread View on postgr.es
From: Alvaro Herrera <alvherre@alvh.no-ip.org>
To: Andres Freund <andres@anarazel.de>
Cc: Michael Paquier <michael@paquier.xyz>, Andrey Borodin <x4mmm@yandex-team.ru>, Robert Haas <robertmhaas@gmail.com>, Peter Geoghegan <pg@bowt.ie>, Петър Славов <pet.slavov@gmail.com>, PostgreSQL mailing lists <pgsql-bugs@lists.postgresql.org>
Date: 2022-05-31T17:35:58Z
Lists: pgsql-bugs 
On 2022-May-30, Andres Freund wrote:

> On 2022-05-30 22:40:39 +0200, Alvaro Herrera wrote:

> > c98763bf51bf Avoid spurious waits in concurrent indexing
> > f9900df5f949 Avoid spurious wait in concurrent reindex
> > d9d076222f5b VACUUM: ignore indexing operations with CONCURRENTLY
> 
> That looks right, yes.

Sorry, after going through these commits again, this is a
misunderstanding on my part.  The first two commits need not be
reverted; only the third one does, which is the one that changes the way
the system-wide Xmin is determined (the commit message says it affects
VACUUM, but naturally it affects HOT pruning as well.)

The point of the first two changes is to remove snapshot waits in CIC
and RC.  Before those commits, each CIC would wait for all other CICs
and RCs in the system.  AFAICS this is unrelated to the HOT-pruning bug.

-- 
Álvaro Herrera         PostgreSQL Developer  —  https://www.EnterpriseDB.com/
"Para tener más hay que desear menos"

Commits

  1. Revert changes to CONCURRENTLY that "sped up" Xmin advance

  2. Fix possible HOT corruption when RECENTLY_DEAD changes to DEAD while pruning.