Horton
Menu

Re: Vacuum statistics

Alena Rybakina <lena.ribackina@yandex.ru>

View thread
From: Alena Rybakina <lena.ribackina@yandex.ru>
To: Andrei Zubkov <zubkov@moonset.ru>
Cc: Andrey Borodin <x4mmm@yandex-team.ru>, Andrei Lepikhov <lepihov@gmail.com>, pgsql-hackers <pgsql-hackers@postgresql.org>, Alexander Korotkov <aekorotkov@gmail.com>, Amit Kapila <amit.kapila16@gmail.com>, Jim Nasby <jnasby@upgrade.com>, Bertrand Drouvot <bertranddrouvot.pg@gmail.com>, Kirill Reshke <reshkekirill@gmail.com>, Masahiko Sawada <sawada.mshk@gmail.com>, Melanie Plageman <melanieplageman@gmail.com>, jian he <jian.universality@gmail.com>, Sami Imseih <samimseih@gmail.com>, vignesh C <vignesh21@gmail.com>, Ilia Evdokimov <ilya.evdokimov@tantorlabs.com>
Date: 2026-03-17T16:38:41Z
Lists: pgsql-hackers

Attachments

Hi!

On 17.03.2026 18:27, Andrei Zubkov wrote:
> Hi, Alena!
>
> I have some thoughts about your descriptions in the docs (V34). It looks
> clearer now, but in seems to me it still contains some inaccuracy..
>
> 1. Naming: The meaning of 'visible/frozen_pages_cleared' seems to me as
> cleared pages rather then cleared marks... Maybe it should be
> 'visible_page_marks_cleared'?
I agree, this improves clarity. I've renamed it as proposed.
> 2. Mention of a <command>VACUUM</command> in the docs may be understood
> as related to manual VACUUM command only. However, autovacuum is
> accounted as well.. I think we can use just term 'vacuum' here as a
> facility rather than command.
>
Fixed. Clarified that this applies to both manual 
<command>VACUUM</command> and autovacuum.

I also added an additional test scenario where one process holds a
transaction open while another process deletes tuples. We expect that
the all-visible and all-frozen flags, previously set by VACUUM, are
cleared only after the deleting transaction commits and the changes
become visible.

-----------
Best regards,
Alena Rybakina

Commits

Same data as JSON: GET /api/v1/messages/:b64id/commits the thread's linked commits as JSON, with link sources. API reference →

  1. Add relallfrozen to pg_class

  2. Move wal_buffers_full from PgStat_PendingWalStats to WalUsage