Horton

RE: parallel vacuum comments

Zhijie Hou (Fujitsu) <houzj.fnst@fujitsu.com>

View thread View on postgr.es
From: "houzj.fnst@fujitsu.com" <houzj.fnst@fujitsu.com>
To: Masahiko Sawada <sawada.mshk@gmail.com>
Cc: PostgreSQL-development <pgsql-hackers@postgresql.org>, Amit Kapila <amit.kapila16@gmail.com>, Peter Geoghegan <pg@bowt.ie>, Andres Freund <andres@anarazel.de>
Date: 2021-11-16T02:38:29Z
Lists: pgsql-hackers 
On Thur, Nov 11, 2021 10:41 AM Masahiko Sawada <sawada.mshk@gmail.com> wrote:
> I've attached a draft patch that refactors parallel vacuum and separates
> parallel-vacuum-related code to new file vacuumparallel.c.
> After discussion, I'll divide the patch into logical chunks.

Hi.

I noticed few minor issues in the patch.

1)
+		/*
+		 * Parallel unsafe indexes can be processed only by leader (these are
+		 * processed in lazy_serial_process_indexes() by leader.
+		 */

It seems the function name in the comments should be serial_vacuum_unsafe_indexes

2)
+		stats->parallel_workers_can_process =
+			index_parallel_vacuum_is_safe(pvc->indrels[i],
+										  pvc->num_index_scans,
+										  bulkdel);

The function index_parallel_vacuum_is_safe also return false for the
index < min_parallel_index_scan_size cutoff which seems parallel safe. So,
maybe we can rename the function to xxx_worker_can_process() ?

Best regards,
Hou zj

Commits

  1. Move parallel vacuum code to vacuumparallel.c.

  2. Move index vacuum routines to vacuum.c.

  3. Improve parallel vacuum implementation.

  4. Fix parallel amvacuumcleanup safety bug.

  5. Simplify state managed by VACUUM.