Re: pgstat_report_activity() and parallel CREATE INDEX (was: Parallel index creation & pg_stat_activity)

Peter Geoghegan <pg@bowt.ie>

From: Peter Geoghegan <pg@bowt.ie>

To: Noah Misch <noah@leadboat.com>

Cc: Andres Freund <andres@anarazel.de>, Phil Florent <philflorent@hotmail.com>, PostgreSQL Hackers <pgsql-hackers@postgresql.org>

Date: 2020-10-17T15:39:35Z

Lists: pgsql-hackers

Commits

Same data as JSON: GET /api/v1/messages/:b64id/commits the thread's linked commits as JSON, with link sources. API reference →

Set debug_query_string in worker_spi.
- d2246cde825e 14.0 landed
Reproduce debug_query_string==NULL on parallel workers.
- 045466675820 11.10 landed
- 741b84e9f747 12.5 landed
- ab2e2ce46668 13.1 landed
- f90e80b91383 14.0 landed
Logical replication support for TRUNCATE
- 039eb6e92f20 11.0 cited
Call pgstat_report_activity() in parallel CREATE INDEX workers.
- 7de4a1bcc56f 11.0 landed

On Tue, Oct 13, 2020 at 7:26 PM Noah Misch <noah@leadboat.com> wrote:
> 1. Disable parallelism for the index build under ExecuteTruncateGuts().
>    Nobody will mourn a performance loss from declining parallelism for an
>    empty index, but I feel like this is fixing in the wrong place.
> 2. Make _bt_begin_parallel() and begin_parallel_vacuum() recognize the
>    debug_query_string==NULL case and reproduce it on the worker.
> 3. Require bgworkers to set debug_query_string before entering code of vacuum,
>    truncate, etc.  Logical replication might synthesize a DDL statement, or it
>    might just use a constant string.
>
> I tend to prefer (2), but (3) isn't bad.  Opinions?

I also prefer 2.

Thanks
-- 
Peter Geoghegan