Horton

Re: wake up logical workers after ALTER SUBSCRIPTION

Kyotaro Horiguchi <horikyota.ntt@gmail.com>

View thread View on postgr.es
From: Kyotaro Horiguchi <horikyota.ntt@gmail.com>
To: nathandbossart@gmail.com
Cc: tgl@sss.pgh.pa.us, houzj.fnst@fujitsu.com, amit.kapila16@gmail.com, m.melihmutlu@gmail.com, thomas.munro@gmail.com, kuroda.hayato@fujitsu.com, pgsql-hackers@postgresql.org
Date: 2023-01-25T07:12:00Z
Lists: pgsql-hackers 
At Tue, 24 Jan 2023 10:42:17 -0800, Nathan Bossart <nathandbossart@gmail.com> wrote in 
> On Tue, Jan 24, 2023 at 01:13:55PM -0500, Tom Lane wrote:
> > Either that comment needs to be rewritten or we need to invent some
> > more macros.
> 
> Here is a first attempt at a patch.  I scanned through all the existing
> uses of InvalidDsaPointer and DSM_HANDLE_INVALID and didn't notice anything
> else that needed adjusting.

There seems to be two cases for DSA_HANDLE_INVALID in dsa_get_handle
and dsa_attach_in_place, one of which is Assert(), though.

regards.

-- 
Kyotaro Horiguchi
NTT Open Source Software Center

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.