Horton

Re: Fixing a couple of buglets in how VACUUM sets visibility map bits

Peter Geoghegan <pg@bowt.ie>

View thread View on postgr.es
From: Peter Geoghegan <pg@bowt.ie>
To: PostgreSQL Hackers <pgsql-hackers@lists.postgresql.org>
Cc: Robert Haas <robertmhaas@gmail.com>
Date: 2023-01-08T22:39:19Z
Lists: pgsql-hackers

Attachments

On Mon, Jan 2, 2023 at 10:31 AM Peter Geoghegan <pg@bowt.ie> wrote:
> Would be helpful if I could get a +1 on
> v1-0002-Never-just-set-the-all-frozen-bit-in-VM.patch, which is
> somewhat more substantial than the others.

There has been no response on this thread for over a full week at this
point. I'm CC'ing Robert now, since the bug is from his commit
a892234f83.

Attached revision of the "don't unset all-visible bit while unsetting
all-frozen bit" patch adds some assertions that verify that
visibility_cutoff_xid is InvalidTransactionId as expected when we go
to set any page all-frozen in the VM. It also broadens an existing
nearby test for corruption, which gives us some chance of detecting
and repairing corruption of this sort that might have slipped in in
the field.

My current plan is to commit something like this in another week or
so, barring any objections.

-- 
Peter Geoghegan

Commits

  1. Tighten up VACUUM's approach to setting VM bits.

  2. Make lazy_vacuum_heap_rel match lazy_scan_heap.

  3. vacuumlazy.c: Tweak local variable name.

  4. vacuumlazy.c: Save get_database_name() in vacrel.

  5. Avoid special XID snapshotConflictHorizon values.

  6. Change the format of the VM fork to add a second bit per page.