Re: Concurrency bug in amcheck

Peter Geoghegan <pg@bowt.ie>

From: Peter Geoghegan <pg@bowt.ie>
To: Alexander Korotkov <aekorotkov@gmail.com>
Cc: pgsql-hackers <pgsql-hackers@postgresql.org>, Michail Nikolaev <michail.nikolaev@gmail.com>
Date: 2020-08-04T22:58:30Z
Lists: pgsql-hackers
Hi Alexander,

On Tue, Aug 4, 2020 at 7:27 AM Alexander Korotkov <aekorotkov@gmail.com> wrote:
> Thank you for your reminder.   Revised patch is attached.  Now, the contents of deleted btree pages isn't masked.  I've checked that installcheck passes with wal_consistency_checking='Btree'.  I'm going to push this if no objections.

This looks good to me. One small thing, though: maybe the comments
should not say anything about the REDO routine -- that seems like a
case of "the tail wagging the dog" to me. Perhaps say something like:

"Remove the last pivot tuple on the page.  This keeps things simple
for WAL consistency checking."

(Just a suggestion.)

Thanks!
-- 
Peter Geoghegan



Commits

  1. Remove btree page items after page unlink

  2. Fix amcheck for page checks concurrent to replay of btree page deletion