Horton

Re: wake up logical workers after ALTER SUBSCRIPTION

Nathan Bossart <nathandbossart@gmail.com>

View thread View on postgr.es
From: Nathan Bossart <nathandbossart@gmail.com>
To: Amit Kapila <amit.kapila16@gmail.com>
Cc: Melih Mutlu <m.melihmutlu@gmail.com>, Thomas Munro <thomas.munro@gmail.com>, "Hayato Kuroda (Fujitsu)" <kuroda.hayato@fujitsu.com>, "pgsql-hackers@postgresql.org" <pgsql-hackers@postgresql.org>
Date: 2023-01-03T18:21:55Z
Lists: pgsql-hackers 
On Tue, Jan 03, 2023 at 11:43:59AM +0530, Amit Kapila wrote:
> On Wed, Dec 7, 2022 at 11:42 PM Nathan Bossart <nathandbossart@gmail.com> wrote:
>> After sleeping on this, I think we can do better.  IIUC we can simply check
>> for AllTablesyncsReady() at the end of process_syncing_tables_for_apply()
>> and wake up the logical replication workers (which should just consiѕt of
>> setting the current process's latch) if we are ready for two_phase mode.
> 
> How just waking up will help with two_phase mode? For that, we need to
> restart the apply worker as we are doing at the beginning of
> process_syncing_tables_for_apply().

Right.  IIRC waking up causes the apply worker to immediately call
process_syncing_tables_for_apply() again, which will then proc_exit(0) as
appropriate.  It might be possible to move the restart logic to the end of
process_syncing_tables_for_apply() to avoid this extra wakeup.  WDYT?

-- 
Nathan Bossart
Amazon Web Services: https://aws.amazon.com

Commits

  1. Avoid type cheats for invalid dsa_handles and dshash_table_handles.

  2. Track logrep apply workers' last start times to avoid useless waits.

  3. Check for two_phase change at end of process_syncing_tables_for_apply.

  4. Wake up a subscription's replication worker processes after DDL.