| From: | Julien Rouhaud <rjuju123(at)gmail(dot)com> |
|---|---|
| To: | Michael Paquier <michael(at)paquier(dot)xyz> |
| Cc: | Postgres hackers <pgsql-hackers(at)lists(dot)postgresql(dot)org> |
| Subject: | Re: Ignore 2PC transaction GIDs in query jumbling |
| Date: | 2023-08-01 01:28:08 |
| Message-ID: | 20230801012808.px3pbet4qk7l3bwk@jrouhaud |
| Views: | Raw Message | Whole Thread | Download mbox | Resend email |
| Thread: | |
| Lists: | pgsql-hackers |
Hi,
On Tue, Aug 01, 2023 at 09:38:14AM +0900, Michael Paquier wrote:
>
> 31de7e6 has silenced savepoint names in the query jumbling, and
> something similar can be done for 2PC transactions once the GID is
> ignored in TransactionStmt. This leads to the following grouping in
> pg_stat_statements, for instance, which is something that matters with
> workloads that heavily rely on 2PC:
> COMMIT PREPARED $1
> PREPARE TRANSACTION $1
> ROLLBACK PREPARED $1
Having an application relying on 2pc leads to pg_stat_statements being
virtually unusable on the whole instance, so +1 for the patch.
FTR we had to entirely ignore all those statements in powa years ago to try to
make the tool usable in such case for some users who where using 2pc, it would
be nice to be able to track them back for pg16+.
The patch LGTM.
| From | Date | Subject | |
|---|---|---|---|
| Next Message | Peter Smith | 2023-08-01 01:35:35 | Simplify some logical replication worker type checking |
| Previous Message | Andres Freund | 2023-08-01 01:25:30 | Re: Performance degradation on concurrent COPY into a single relation in PG16. |