From: Tom Lane <tgl@sss.pgh.pa.us>
To: feliphepozzer@gmail.com
cc: pgsql-bugs@lists.postgresql.org, Richard Guo <guofenglinux@gmail.com>
Subject: Re: BUG #18526: 'UPDATE' inconsistency using index scan with 'NOT
 EXIST' after upgrading to PostgreSQL 16
In-reply-to: <18526-bf89f27cb20d8a18@postgresql.org>
References: <18526-bf89f27cb20d8a18@postgresql.org>
Comments: In-reply-to PG Bug reporting form <noreply@postgresql.org>
	message dated "Fri, 28 Jun 2024 17:24:32 -0000"
MIME-Version: 1.0
Content-Type: text/plain; charset="us-ascii"
Content-ID: <2002049.1719602033.1@sss.pgh.pa.us>
Content-Transfer-Encoding: quoted-printable
Date: Fri, 28 Jun 2024 15:13:53 -0400
Message-ID: <2002050.1719602033@sss.pgh.pa.us>
Archived-At: 
 <https://www.postgresql.org/message-id/2002050.1719602033%40sss.pgh.pa.us>
Precedence: bulk

PG Bug reporting form <noreply@postgresql.org> writes:
> Expected Result: All records that meet the criteria should be updated
> regardless of using index scan or seq scan.
> Actual Result: When PostgreSQL 16 uses index scan, it fails to find and
> update all records. By changing random_page_cost to a value that forces =
the
> use of seq scan, all records are updated correctly.

I see that the problematic plan involves a Merge Right Anti Join step,
which makes me guess that this is the same bug recently reported in

https://www.postgresql.org/message-id/flat/18522-c7a8956126afdfd0%40postgr=
esql.org

Are you in a position to try the patch posted in that thread?

			regards, tom lane