Horton

Re: XID formatting and SLRU refactorings (was: Add 64-bit XIDs into PostgreSQL 15)

Aleksander Alekseev <aleksander@timescale.com>

View thread View on postgr.es
From: Aleksander Alekseev <aleksander@timescale.com>
To: Postgres hackers <pgsql-hackers@lists.postgresql.org>
Cc: Pavel Borisov <pashkin.elfe@gmail.com>, Maxim Orlov <orlovmg@gmail.com>, Japin Li <japinli@hotmail.com>, Alexander Korotkov <aekorotkov@gmail.com>, Kyotaro Horiguchi <horikyota.ntt@gmail.com>, Ilya Anfimov <ilan@tzirechnoy.com>, Peter Eisentraut <peter.eisentraut@enterprisedb.com>
Date: 2022-03-22T13:21:10Z
Lists: pgsql-hackers 
Hi hackers,

> Here is v23. As was suggested by Alexander above, I've changed the order of the patches and improved the commit message. Now, SLRU patch is the first.

Many thanks!

> There were proposals to make use of SLRU pages numbers that are in fact unsigned and change from int to uint64. I fully support this, but I'm not sure this big SLRU refactoring should be done in this patchset.

If it takes a lot of effort and doesn't bring us any closer to 64-bit
XIDs, I suggest not doing this in v23-0001. I can invest some time
into this refactoring in April and create a separate CF entry, if
someone will second the idea.

> In general, I consider this patchset is ready to commit. It would be great to deliver it in PG15.

+1.

v23-0002 seems to have two extra sentences in the commit message that
are outdated, but this is a minor issue. The commit message should be:

"""
Replace the %u formatting string for XIDs with %llu and cast to
unsigned long long. While actually XIDs are still 32 bit, this patch
completely supports both 32 and 64 bit.
"""

Since Peter expressed some concerns regarding v23-0002, maybe we
should discuss it a bit more. Although personally I doubt that we can
do much better than that, and as I recall this particular change was
explicitly requested by several people.

-- 
Best regards,
Aleksander Alekseev

Commits

  1. Fix more holes with SLRU code in need of int64 for segment numbers

  2. Use more consistently int64 for page numbers in SLRU-related code

  3. Improve comments in slru.{c,h} about segment name format

  4. Reorder actions in ProcArrayApplyRecoveryInfo()

  5. Fix typo in 5a1dfde8334b

  6. Fix warning due non-standard inline declaration in 4ed8f0913bfdb5f355

  7. Add SLRU tests for 64-bit page case

  8. Make use FullTransactionId in 2PC filenames

  9. Use larger segment file names for pg_notify

  10. Index SLRUs by 64-bit integers rather than by 32-bit integers

  11. Keep track of transaction commit timestamps

  12. Handle 5-char filenames in SlruScanDirectory