Horton
Menu

Re: 64-bit XIDs in deleted nbtree pages

Peter Geoghegan <pg@bowt.ie>

View thread
From: Peter Geoghegan <pg@bowt.ie>
To: Masahiko Sawada <sawada.mshk@gmail.com>
Cc: PostgreSQL Hackers <pgsql-hackers@lists.postgresql.org>, Heikki Linnakangas <hlinnaka@iki.fi>
Date: 2021-02-11T03:50:40Z
Lists: pgsql-hackers 
On Wed, Feb 10, 2021 at 7:10 PM Peter Geoghegan <pg@bowt.ie> wrote:
> Attached is v3 of the index. I'll describe the changes I made in more
> detail in my response to your points below.

I forget to mention that v3 adds several assertions like this one:

Assert(!_bt_page_recyclable(BufferGetPage(buf)));

These appear at a few key points inside generic routines like
_bt_getbuf(). The overall effect is that every nbtree buffer access
(with the exception of buffer accesses by VACUUM) will make sure that
the page that they're about to access is not recyclable (a page that
an index scan lands on might be half-dead or deleted, but it had
better not be recyclable).

This can probably catch problems with recycling pages too early, such
as the problem fixed by commit d3abbbeb back in 2012. Any similar bugs
in this area that may appear in the future can be expected to be very
subtle, for a few reasons. For one, a page can be recyclable but not
yet entered into the FSM by VACUUM for a long time. (I could go on.)

The assertions dramatically improve our chances of catching problems
like that early.

-- 
Peter Geoghegan

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.