| From: | Ajin Cherian <itsajin(at)gmail(dot)com> |
|---|---|
| To: | Amit Kapila <amit(dot)kapila16(at)gmail(dot)com> |
| Cc: | vignesh C <vignesh21(at)gmail(dot)com>, Alexander Lakhin <exclusion(at)gmail(dot)com>, pgsql-hackers <pgsql-hackers(at)postgresql(dot)org> |
| Subject: | Re: 024_add_drop_pub.pl might fail due to deadlock |
| Date: | 2025-07-22 11:58:32 |
| Message-ID: | CAFPTHDYu38w1EYU_MD+29-mdXpV2F9gCzwBYAxuKVFm6auLYBA@mail.gmail.com |
| Views: | Whole Thread | Raw Message | Download mbox | Resend email |
| Thread: | |
| Lists: | pgsql-hackers |
> locked. I've made a new version of the patch on PG_15.
I've made a similar fix on HEAD just so that the code is now
consistent. I don't think the similar problem (deadlock between two
different subscriptions trying to drop tracking origin) occurs on
HEAD with my previous patch, as the way origins are dropped are
different on HEAD. On HEAD, while dropping origin, a RowExclusiveLock
lock is taken on ReplicationOriginRelationId and then an
AccessExclusiveLock is taken on the particular origin. Since the
particular origin will be different on different subscriptions, the
similar deadlock will not happen. But just to keep code consistent, I
have made a similar fix.
regards,
Ajin Cherian
Fujitsu Australia
| Attachment | Content-Type | Size |
|---|---|---|
| HEAD-v3-0001-Fix-a-possible-deadlock-during-ALTER-SUBSCRIPTION.patch | application/octet-stream | 7.5 KB |
| From | Date | Subject | |
|---|---|---|---|
| Next Message | Andrei Lepikhov | 2025-07-22 12:04:31 | Re: track generic and custom plans in pg_stat_statements |
| Previous Message | Matheus Alcantara | 2025-07-22 11:55:56 | Re: Proposal: QUALIFY clause |