Regress test might fail due to deadlock between domain and alter_table

From: Alexander Lakhin <exclusion(at)gmail(dot)com>
To: pgsql-hackers <pgsql-hackers(at)postgresql(dot)org>
Subject: Regress test might fail due to deadlock between domain and alter_table
Date: 2026-10-07 20:00:00
Message-ID: b036460a-f25f-4043-a5cb-6e77a9fd53eb@gmail.com
Views: Whole Thread | Raw Message | Download mbox | Resend email
Thread:
Lists: pgsql-hackers

Hello hackers,

I happened to observe an interesting failure of `make check` being
executed concurrently against multiple databases:
diff -U3 --minimal .../regress_input_27/expected/domain.out .../regress_output_27/results/domain.out
--- .../expected/domain.out    2026-10-02 08:03:59.474274001 +0000
+++ .../results/domain.out    2026-10-02 08:26:49.225597020 +0000
@@ -1337,11 +1337,7 @@
             FROM information_schema.domain_constraints
             WHERE domain_name IN ('con', 'dom', 'pos_int', 'things'))
   ORDER BY constraint_name;
- constraint_catalog | constraint_schema | constraint_name  | check_clause
---------------------+-------------------+------------------+-------------------
- regress027         | public            | con_check        | (VALUE > 0)
- regress027         | public            | meow             | (VALUE < 11)
- regress027         | public            | pos_int_check    | (VALUE > 0)
- regress027         | public            | pos_int_not_null | VALUE IS NOT NULL
-(4 rows)
-
+ERROR:  deadlock detected
+DETAIL:  Process 322216 waits for AccessShareLock on relation 1197300 of database 16411; blocked by process 322217.
+Process 322217 waits for AccessExclusiveLock on relation 1197314 of database 16411; blocked by process 322216.
+HINT:  See server log for query details.

I've managed to reproduce this with the attached patch:
# diff -U3 .../src/test/regress/expected/alter_table.out .../src/test/regress/results/alter_table.out
# --- .../src/test/regress/expected/alter_table.out       2026-10-07 12:53:24.843538437 -0400
# +++ .../src/test/regress/results/alter_table.out        2026-10-07 14:58:08.602062642 -0400
# @@ -4808,7 +4808,10 @@
#  (1 row)
#
#  alter table attbl alter column a type numeric;  -- someday this should work
# -ERROR:  cannot alter table "attbl" because column "atref.b" uses its row type
# +ERROR:  deadlock detected
# +DETAIL:  Process 2884690 waits for AccessExclusiveLock on relation 45440 of database 16384; blocked by process 2884685.
# +Process 2884685 waits for AccessShareLock on relation 45437 of database 16384; blocked by process 2884690.
# +HINT:  See server log for query details.
#  alter table atref drop constraint atref_b_check;
#  create statistics atref_stat on ((b).a is not null) from atref;
#  alter table attbl alter column a type numeric;  -- someday this should work

That is, the deadlock occurs between
"SELECT * FROM information_schema.check_constraint" / pg_get_constraintdef()
and
"alter table attbl alter column a type numeric;" / ATPostAlterTypeCleanup().

Reproduced starting from 8319e5cb5.

Best regards,
Alexander

Attachment Content-Type Size
reproduce-deadlock.patch text/x-patch 3.8 KB

Responses

Browse pgsql-hackers by date

  From Date Subject
Next Message Osama Abdul Qader 2026-10-07 20:24:50 Re: Regress test might fail due to deadlock between domain and alter_table
Previous Message Ayoub Kazar 2026-10-07 19:53:27 Re: [PROPOSAL] Expand OR clauses in joins to UNION ALL paths