Horton

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

Michael Paquier <michael@paquier.xyz>

View thread View on postgr.es
From: Michael Paquier <michael@paquier.xyz>
To: Alvaro Herrera <alvherre@alvh.no-ip.org>
Cc: Andres Freund <andres@anarazel.de>, 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: 2022-05-31T12:03:17Z
Lists: pgsql-bugs 
On Tue, May 31, 2022 at 11:19:01AM +0200, Alvaro Herrera wrote:
> Hmm, I wonder if it's possible to use an index ON tab (pg_sleep(100ms))
> or something like that (perhaps even use an expresion involving an
> advisory lock acquisition, which is being held by the other session), to
> widen the window.

I have not considered the use of expressions to enforce a lock at this
stage.  I'll give it a shot, even if it comes to the abuse of an
immutable function with volatile internals :)

> The stop events patch that A. Korotkov posted [0] would likely be very
> useful here.
> 
> [0] https://postgr.es/m/CAPpHfdtSEOHX8dSk9Qp+Z++i4BGQoffKip6JDWngEA+g7Z-XmQ@mail.gmail.com

Thanks for the reference.  I've wanted to resurrect this stuff,
actually, after dealing with 6dced63 to be able to have a stop point
in the middle of a restart point.
--
Michael

Commits

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

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