From: | Simon Riggs <simon(dot)riggs(at)enterprisedb(dot)com> |
---|---|
To: | PostgreSQL Hackers <pgsql-hackers(at)postgresql(dot)org>, Alvaro Herrera <alvherre(at)alvh(dot)no-ip(dot)org> |
Subject: | SKIP LOCKED assert triggered |
Date: | 2021-11-12 16:55:17 |
Message-ID: | CANbhV-FeEwMnN8yuMyss7if1ZKjOKfjcgqB26n8pqu1e=q0ebg@mail.gmail.com |
Views: | Raw Message | Whole Thread | Download mbox | Resend email |
Thread: | |
Lists: | pgsql-hackers |
The combination of these two statements in a transaction hits an
Assert in heapam.c at line 4770 on REL_14_STABLE
BEGIN;
SELECT * FROM queue LIMIT 1 FOR UPDATE SKIP LOCKED;
...
UPDATE queue SET status = 'UPDATED' WHERE id = :id;
COMMIT;
pgbench reliably finds this, running from inside a PL/pgSQL function.
Alvaro suggests we just ignore the Assert, which on testing appears to
be the right approach. Patch attached, which solves it for me.
There is no formal test that does lock then update, so I have
attempted to create one, but this has not successfully reproduced it
(attached anyway), but this code is different from the pgbench test.
--
Simon Riggs http://www.EnterpriseDB.com/
Attachment | Content-Type | Size |
---|---|---|
skip_locked_assert.v1.patch | application/octet-stream | 629 bytes |
skip_locked_then_update_test.v1.patch | application/octet-stream | 54.5 KB |
From | Date | Subject | |
---|---|---|---|
Next Message | Fabrice Chapuis | 2021-11-12 16:57:41 | Re: Logical replication timeout problem |
Previous Message | Tom Lane | 2021-11-12 16:54:20 | Re: [PATCH] postgres_fdw: suppress explicit casts in text:text comparisons (was: column option to override foreign types) |