MIME-Version: 1.0
References: <4d301501-9985-4e68-b1c1-2360c251048c@cloud.gatewaynet.com> <54d352ad-3f64-4492-aaaf-1ebaeac5471c@aklaver.com>
In-Reply-To: <54d352ad-3f64-4492-aaaf-1ebaeac5471c@aklaver.com>
From: Paul A Jungwirth <pj@illuminatedcomputing.com>
Date: Wed, 20 Aug 2025 09:49:31 -0500
Message-ID: <CA+renyU=3z2J_yLaz1BFk9sHeO1Gnu=KMof5vcb-diW1_JZggQ@mail.gmail.com>
Subject: Re: pg_upgrade from 18beta1 -> 18beta3 - problem with btree_gist
 contrib module / extension
To: Adrian Klaver <adrian.klaver@aklaver.com>
Cc: Achilleas Mantzios <a.mantzios@cloud.gatewaynet.com>, 
	"pgsql-general@lists.postgresql.org" <pgsql-general@lists.postgresql.org>
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable
Archived-At: <https://www.postgresql.org/message-id/CA%2BrenyU%3D3z2J_yLaz1BFk9sHeO1Gnu%3DKMof5vcb-diW1_JZggQ%40mail.gmail.com>
Precedence: bulk

On Wed, Aug 20, 2025 at 9:39=E2=80=AFAM Adrian Klaver <adrian.klaver@aklave=
r.com> wrote:
>
> On 8/20/25 05:50, Achilleas Mantzios wrote:
> > pg_restore: from TOC entry 4295; 1255 596569951 FUNCTION
> > gist_stratnum_btree(integer) postgres
> > pg_restore: error: could not execute query: ERROR:  could not find
> > function "gist_stratnum_btree" in file "/usr/local/pgsql/lib/btree_gist=
.so"
> > Command was: CREATE FUNCTION "public"."gist_stratnum_btree"(integer)
> > RETURNS smallint
> >     LANGUAGE "c" IMMUTABLE STRICT PARALLEL SAFE
> >     AS '$libdir/btree_gist', 'gist_stratnum_btree';
>
> I can not find gist_stratnum_btree in:
>
> https://www.postgresql.org/docs/18/btree-gist.html
>
> or in the source.
>
> How did it end up in the database?

gist_stratnum_btree was in beta1 but was renamed to
gist_translate_cmptype_btree by
32edf732e8dc9eb3e7a923aeb67d49246744a20a.

> > The solution was to somehow restart 18beta1, drop btree_gist and all it=
s
> > dependent constraints / indexes / objects , pg_upgrade and finally re-
> > create extension and dependent objects.
>
> Something more then DROP EXTENSION/CREATE EXTENSION btree_gist?

That seems like the easiest fix to me. But if you've built indexes
based on those opclasses then more work is required, as here.

I don't know what pg_upgrade tries to support for moving from one beta
release to another. Is this something we should try to fix?

Yours,

--=20
Paul              ~{:-)
pj@illuminatedcomputing.com