Horton

Re: old_snapshot_threshold bottleneck on replica

Robert Haas <robertmhaas@gmail.com>

View thread View on postgr.es
From: Robert Haas <robertmhaas@gmail.com>
To: Maxim Orlov <orlovmg@gmail.com>
Cc: Postgres hackers <pgsql-hackers@lists.postgresql.org>
Date: 2023-01-25T13:52:28Z
Lists: pgsql-hackers 
On Wed, Jan 25, 2023 at 3:52 AM Maxim Orlov <orlovmg@gmail.com> wrote:
> But in TransactionIdLimitedForOldSnapshots these variable is using conjointly. So, I'm not
> sure, is it completely safe to remove mutex.

Well, that's something we - and ideally you, as the patch author -
need to analyze and figure out. We can't just take a shot and hope for
the best.

> Actually, removing mutex and switch to atomics
> was my first choice. I've run all the tests and no problems were found

Unfortunately, that kind of testing is not very likely to find a
subtle synchronization problem. That's why a theoretical analysis is
so important.

-- 
Robert Haas
EDB: http://www.enterprisedb.com

Commits

  1. Remove some more "snapshot too old" vestiges.

  2. Remove the "snapshot too old" feature.