Horton

RE: Is this a problem in GenericXLogFinish()?

Hayato Kuroda (Fujitsu) <kuroda.hayato@fujitsu.com>

View thread View on postgr.es
From: "Hayato Kuroda (Fujitsu)" <kuroda.hayato@fujitsu.com>
To: 'Amit Kapila' <amit.kapila16@gmail.com>, Michael Paquier <michael@paquier.xyz>
Cc: Robert Haas <robertmhaas@gmail.com>, Jeff Davis <pgsql@j-davis.com>, Alexander Lakhin <exclusion@gmail.com>, Heikki Linnakangas <hlinnaka@iki.fi>, "pgsql-hackers@postgresql.org" <pgsql-hackers@postgresql.org>
Date: 2023-11-10T03:47:30Z
Lists: pgsql-hackers

Attachments

Dear hackers,

> Next we should add some test codes. I will continue considering but please post
> anything
> If you have idea.

And I did, PSA the patch. This patch adds two parts in hash_index.sql.

In the first part, the primary bucket page is filled by live tuples and some overflow
pages are by dead ones. This leads removal of overflow pages without moving tuples.
HEAD will crash if this were executed, which is already reported on hackers.

The second one tests a case (ntups == 0 && is_prim_bucket_same_wrt == false &&
is_prev_bucket_same_wrt == true), which is never done before.



Also, I measured the execution time of before/after patching. Below is a madian
for 9 measurements.

BEFORE -> AFTER
647 ms -> 710 ms

This means that the execution time increased -10%, it will not affect so much.

How do you think?

Best Regards,
Hayato Kuroda
FUJITSU LIMITED

Commits

  1. Fix inconsistency with replay of hash squeeze record for clean buffers

  2. Set LSN for wbuf in _hash_freeovflpage() iff wbuf is modified.

  3. Fix an uninitialized access in hash_xlog_squeeze_page().

  4. Use REGBUF_NO_CHANGE at one more place in the hash index.

  5. Assert that buffers are marked dirty before XLogRegisterBuffer().

  6. Fix bug in GenericXLogFinish().

  7. Improve hash index bucket split behavior.