Horton

Re: use CREATE DATABASE STRATEGY = FILE_COPY in pg_upgrade

Nathan Bossart <nathandbossart@gmail.com>

View thread View on postgr.es
From: Nathan Bossart <nathandbossart@gmail.com>
To: Matthias van de Meent <boekewurm+postgres@gmail.com>
Cc: Ranier Vilela <ranier.vf@gmail.com>, pgsql-hackers@postgresql.org
Date: 2024-06-05T17:45:48Z
Lists: pgsql-hackers 
On Wed, Jun 05, 2024 at 07:28:42PM +0200, Matthias van de Meent wrote:
> As for "on my laptop", that sounds very reasonable, but could you
> check the performance on systems with larger shared buffer
> configurations? I'd imagine (but haven't checked) that binary upgrade
> target systems may already be using the shared_buffers from their
> source system, which would cause a severe regression when the
> to-be-upgraded system has large shared buffers. For initdb the
> database size is known in advance and shared_buffers is approximately
> empty, but the same is not (always) true when we're doing binary
> upgrades.

Will do.  FWIW I haven't had much luck improving pg_upgrade times by
adjusting server settings, but I also haven't explored it all that much.

-- 
nathan

Commits

  1. Use CREATE DATABASE ... STRATEGY = FILE_COPY in pg_upgrade.

  2. Use TRUNCATE to preserve relfilenode for pg_largeobject + index.

  3. initdb: When running CREATE DATABASE, use STRATEGY = WAL_COPY.