MIME-Version: 1.0
From: "Colin 't Hart" <colinthart@gmail.com>
Date: Wed, 29 Oct 2025 12:27:44 +0100
Message-ID: <CAMon-aTbZC5k2pLpijk5x+h__LV08DzpoJtHBgZGkOKJNJ91pw@mail.gmail.com>
Subject: Two sequences associated with one identity column
To: pgsql-general@lists.postgresql.org
Content-Type: text/plain; charset="UTF-8"
Archived-At: <https://www.postgresql.org/message-id/CAMon-aTbZC5k2pLpijk5x%2Bh__LV08DzpoJtHBgZGkOKJNJ91pw%40mail.gmail.com>
Precedence: bulk

Hi,

One of my clients has a database in which a single identity column
(called "id" in that table) has two sequences associated with it(!)

Both sequences display

Sequence for identity column: <schema>.<table>.id

when described with \d in psql.


Inserting fails with "ERROR:  more than one owned sequence found", as
does trying to alter the table to drop the identity on that column.


Trying to drop either sequence results in

ERROR:  cannot drop sequence <name> because column id of table <name>
requires it
HINT:  You can drop column id of table <name> instead.

while trying to alter either sequence "owned by none" results in

ERROR:  cannot change ownership of identity sequence
DETAIL:  Sequence "<name>" is linked to table "<name>".


How do we fix this? I presume we need to update the catalog directly
to dissociate one of the sequences and after that drop the orphaned
sequence.


This is in a Postgres 12.22 database that we're trying to upgrade to Postgres 17


Thanks,

Colin