Horton

Re: Add mode column to pg_stat_progress_vacuum

Shinya Kato <shinya11.kato@gmail.com>

View thread
From: Shinya Kato <shinya11.kato@gmail.com>
To: Sami Imseih <samimseih@gmail.com>
Cc: maciek@sakrejda.org, Masahiko Sawada <sawada.mshk@gmail.com>, Nathan Bossart <nathandbossart@gmail.com>, Robert Treat <rob@xzilla.net>, pgsql-hackers <pgsql-hackers@postgresql.org>
Date: 2025-11-20T06:23:32Z
Lists: pgsql-hackers

Attachments

On Thu, Nov 20, 2025 at 10:34 AM Sami Imseih <samimseih@gmail.com> wrote:
>
> Hi,
>
> The patch looks overall good to me. I have a few comments:

Thank you for your review!

> 1/ I believe It is possible for the mode to flip to "failsafe" during the
> operation since lazy_check_wraparound_failsafe() is checked in-flight.
> I think this should be mentioned in the "mode" section of the documentation
> in the "failsafe" description.

+1, I've added the below statement to the v7 patch.
---
The vacuum may start in this mode or switch to it while running; the
value of the mode column may transition from another mode to failsafe
during the operation.
---

> 2/ It feels odd that we do not have parity of the "mode" column in
> pg_stat_progress_analyze. Should we add a "mode" column
> there also that shows if the analyze was triggered_by "manual" or
> "autovacuum"? An autovacuum in a relation can trigger both
> a vacuum and an analyze, and a VACUUM(ANALYZE) will
> also do both.

Thank you for your suggestions. I've created another patch v7-0002
adding a "triggered_by" column to pg_stat_progress_analyze.

-- 
Best regards,
Shinya Kato
NTT OSS Center