Horton

Re: Non-replayable WAL records through overflows and >MaxAllocSize lengths

Michael Paquier <michael@paquier.xyz>

View thread View on postgr.es
From: Michael Paquier <michael@paquier.xyz>
To: Matthias van de Meent <boekewurm+postgres@gmail.com>
Cc: Heikki Linnakangas <hlinnaka@iki.fi>, David Zhang <david.zhang@highgo.ca>, Andres Freund <andres@anarazel.de>, Robert Haas <robertmhaas@gmail.com>, PostgreSQL Hackers <pgsql-hackers@postgresql.org>
Date: 2022-10-05T07:46:25Z
Lists: pgsql-hackers 
Hi Matthias,

On Wed, Jul 27, 2022 at 02:07:05PM +0200, Matthias van de Meent wrote:

My apologies for the time it took me to come back to this thread.
> > + * To accommodate some overhead, hhis MaxXLogRecordSize value allows for
> > s/hhis/this/.
> 
> Will be included in the next update..

v8 fails to apply.  Could you send a rebased version?

As far as I recall the problems with the block image sizes are solved,
but we still have a bit more to do in terms of the overall record
size.  Perhaps there are some parts of the patch you'd like to
revisit?

For now, I have switched the back as waiting on author, and moved it
to the next CF.
--
Michael

Commits

  1. Add more protections in WAL record APIs against overflows

  2. Add overflow protection for block-related data in WAL records

  3. Change internal RelFileNode references to RelFileNumber or RelFileLocator.

  4. Revamp the WAL record format.