BUG #18609: Repeated installcheck failure in test_pg_dump due to existing role

Started by PG Bug reporting formover 1 year ago5 messagesbugs
Jump to latest
#1PG Bug reporting form
noreply@postgresql.org

The following bug has been logged on the website:

Bug reference: 18609
Logged by: Nikita Kalinin
Email address: n.kalinin@postgrespro.ru
PostgreSQL version: 16.4
Operating system: ubuntu 22.04
Description:

Hello!
When running the installcheck a second time, the test_pg_dump fails due to
the following reason:
CREATE ROLE regress_dump_test_role;
+ERROR: role "regress_dump_test_role" already exists
CREATE EXTENSION test_pg_dump;

This issue is present from REL_12_STABLE to REL_16_STABLE. Maybe it's worth
adding DROP ROLE regress_dump_test_role at the end of the test, like in
REL_17_STABLE?

#2Daniel Gustafsson
daniel@yesql.se
In reply to: PG Bug reporting form (#1)
Re: BUG #18609: Repeated installcheck failure in test_pg_dump due to existing role

On 11 Sep 2024, at 08:42, PG Bug reporting form <noreply@postgresql.org> wrote:

The following bug has been logged on the website:

Bug reference: 18609
Logged by: Nikita Kalinin
Email address: n.kalinin@postgrespro.ru
PostgreSQL version: 16.4
Operating system: ubuntu 22.04
Description:

Hello!
When running the installcheck a second time, the test_pg_dump fails due to
the following reason:
CREATE ROLE regress_dump_test_role;
+ERROR: role "regress_dump_test_role" already exists
CREATE EXTENSION test_pg_dump;

This issue is present from REL_12_STABLE to REL_16_STABLE. Maybe it's worth
adding DROP ROLE regress_dump_test_role at the end of the test, like in
REL_17_STABLE?

That would be backpatching 936e3fa3787a51397280c1081587586e83c20399. I haven't
checked if there are complications in doing so but I think it should be quite
straightforward.

--
Daniel Gustafsson

#3Michael Paquier
michael@paquier.xyz
In reply to: Daniel Gustafsson (#2)
Re: BUG #18609: Repeated installcheck failure in test_pg_dump due to existing role

On Wed, Sep 11, 2024 at 10:01:22AM +0200, Daniel Gustafsson wrote:

That would be backpatching 936e3fa3787a51397280c1081587586e83c20399. I haven't
checked if there are complications in doing so but I think it should be quite
straightforward.

As far as I can see it would be enough to add the DROP OWNED BY to
avoid the random notice messages listing the dependencies the role
depends on, followed by the DROP ROLE. FWIW, I am not seeing any
complications (the TAP test recreates the same role name
regress_dump_test_role on its own), so let's just do it.
--
Michael

#4Никита Калинин
n.kalinin@postgrespro.ru
In reply to: Michael Paquier (#3)
Re: BUG #18609: Repeated installcheck failure in test_pg_dump due to existing role

Yes, it will be enough to add DROP OWNED BY.
Perhaps with these changes, we can allow meson builds to run the test_pg_dump test in installcheck mode on the REL_16_STABLE branch?

Show quoted text

On 19 Sep 2024, at 14:44, Michael Paquier <michael@paquier.xyz> wrote:

DROP OWNED BY

#5Michael Paquier
michael@paquier.xyz
In reply to: Никита Калинин (#4)
Re: BUG #18609: Repeated installcheck failure in test_pg_dump due to existing role

On Mon, Sep 23, 2024 at 03:21:13PM +0700, Никита Калинин wrote:

Yes, it will be enough to add DROP OWNED BY.
Perhaps with these changes, we can allow meson builds to run the
test_pg_dump test in installcheck mode on the REL_16_STABLE branch?

Yes, like b0c5b215dace, we should. REL_17_STABLE is frozen this week
for the new major release, but the other branches are not, so I have
applied a cherry-pick of 936e3fa3787a to 12~16, and b0c5b215dace to
16.
--
Michael