Re: REPACK (CONCURRENTLY) can lose data in pg_dump output

From: shihao zhong <zhong950419(at)gmail(dot)com>
To: Sami Imseih <samimseih(dot)pg(at)gmail(dot)com>
Cc: Thom Brown <thom(at)linux(dot)com>, PostgreSQL Hackers <pgsql-hackers(at)lists(dot)postgresql(dot)org>
Subject: Re: REPACK (CONCURRENTLY) can lose data in pg_dump output
Date: 2026-09-25 22:32:30
Message-ID: CAGRkXqQ05fEx7QdfbMtsYu4GxaeAcYRGn0rd0Ot-=ZsSpC0fyA@mail.gmail.com
Views: Whole Thread | Raw Message | Download mbox | Resend email
Thread:
Lists: pgsql-hackers

Hi Thom,

In my understanding, this is an expected beahcior as you mentioned,

A fix is already proposed for v20. 0008 in Antonin's "REPACK
enhancements" series, "Make REPACK (CONCURRENTLY) MVCC-safe", carries
the visibility info over to the new table [1][2]. Jochen hit the same
problem on -bugs with a repeatable read session[3].

[1] https://commitfest.postgresql.org/patch/6957/
[2] https://www.postgresql.org/message-id/224072.1789577949@localhost
[3]
https://www.postgresql.org/message-id/0865e13a-da93-4ad1-8b93-7b54a1628f04@jbitc.de

Regards,
Shihao

In response to

Browse pgsql-hackers by date

  From Date Subject
Next Message Bharath Rupireddy 2026-09-25 22:41:06 Re: [PATCH] Refactor remaining zero-fill relation extensions to use smgrzeroextend()
Previous Message Masahiko Sawada 2026-09-25 22:24:06 Re: Fix "unexpected logical decoding status change" error; from concurrent logical decoding activation