Horton
Menu

Re: 64-bit XIDs in deleted nbtree pages

Masahiko Sawada <sawada.mshk@gmail.com>

View thread
From: Masahiko Sawada <sawada.mshk@gmail.com>
To: Peter Geoghegan <pg@bowt.ie>
Cc: PostgreSQL Hackers <pgsql-hackers@lists.postgresql.org>, Heikki Linnakangas <hlinnaka@iki.fi>
Date: 2021-03-23T15:13:56Z
Lists: pgsql-hackers 
On Mon, Mar 22, 2021 at 7:27 AM Peter Geoghegan <pg@bowt.ie> wrote:
>
> On Wed, Mar 10, 2021 at 5:34 PM Peter Geoghegan <pg@bowt.ie> wrote:
> > Here is another bitrot-fix-only revision, v9. Just the recycling patch again.
>
> I committed the final nbtree page deletion patch just now -- the one
> that attempts to make recycling happen for newly deleted pages. Thanks
> for all your work on patch review, Masahiko!

You're welcome! Those are really good improvements.

By this patch series, btree indexes became like hash indexes in terms
of amvacuumcleanup. We do an index scan at btvacuumcleanup() in the
two cases: metapage upgrading and more than 5%
deleted-but-not-yet-recycled pages. Both cases seem rare cases. So do
we want to disable parallel index cleanup for btree indexes like hash
indexes? That is, remove VACUUM_OPTION_PARALLEL_COND_CLEANUP from
amparallelvacuumoptions. IMO we can live with the current
configuration just in case where the user runs into such rare
situations (especially for the latter case). In most cases, parallel
vacuum workers for index cleanup might exit with no-op but the
side-effect (wasting resources and overhead etc) would not be big. If
we want to enable it only in particular cases, we would need to have
another way for index AM to tell lazy vacuum whether or not to allow a
parallel worker to process the index at that time. What do you think?

I’m not sure we need changes but I think it’s worth discussing here.

Regards,

-- 
Masahiko Sawada
EDB:  https://www.enterprisedb.com/

Commits

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

  1. Recycle nbtree pages deleted during same VACUUM.

  2. VACUUM VERBOSE: Count "newly deleted" index pages.

  3. Use full 64-bit XIDs in deleted nbtree pages.

  4. Fix nbtree cleanup-only VACUUM stats inaccuracies.

  5. Fix undercounting in VACUUM VERBOSE output.

  6. Delete empty pages in each pass during GIST VACUUM.

  7. Use full 64-bit XID for checking if a deleted GiST page is old enough.

  8. Allow VACUUM to be run with index cleanup disabled.

  9. Skip full index scan during cleanup of B-tree indexes when possible

  10. Do index FSM vacuuming sooner.

  11. Avoid early reuse of btree pages, causing incorrect query results.

  12. Clean up API for ambulkdelete/amvacuumcleanup as per today's discussion.