Re: BUG #19893: REPLICA IDENTITY FULL: UPDATE/DELETE skipped when a subscriber-only column has a value

From: shihao zhong <zhong950419(at)gmail(dot)com>
To: Laurenz Albe <laurenz(dot)albe(at)cybertec(dot)at>
Cc: kehan5800(at)gmail(dot)com, pgsql-bugs(at)lists(dot)postgresql(dot)org
Subject: Re: BUG #19893: REPLICA IDENTITY FULL: UPDATE/DELETE skipped when a subscriber-only column has a value
Date: 2026-11-19 05:06:11
Message-ID: CAGRkXqQ-Oc5EXGC0YENCCOXeJEAC041030oFRy6Ld2a-cmxxAQ@mail.gmail.com
Views: Whole Thread | Raw Message | Download mbox | Resend email
Thread:
Lists: pgsql-bugs

Hi Laurenz,

I think this is a bug. The docs suggest adding a column on the
subscriber first, and that alone breaks replication of the table.

Take a table with REPLICA IDENTITY FULL and no primary key. Run
ALTER TABLE t ADD COLUMN c int DEFAULT 0 on the subscriber. Then
update one row and delete another on the publisher. Both changes are
skipped. The two rows stay wrong on the subscriber for good, even
after the publisher adds the same column.

We fixed the similar thing for dropped and generated columns in
b797def5951 and adedf54e659, because the publisher does not send them.
Subscriber-only columns are not sent either.

WDYT?

Thanks,
Shihao

In response to

Browse pgsql-bugs by date

  From Date Subject
Next Message shihao zhong 2026-11-19 05:09:11 Re: BUG #19899: PL/Python backend crashes (SIGSEGV) when converting a column value to a Python object fails
Previous Message TAKATSUKA Haruka 2026-11-19 01:42:03 Re: BUG #19897: Recovery fails after "LOG: consistent recovery state reached"