Oct 2026

1 commit
Refresh autovacuum costs while waiting for parallel workers.
1 Oct 2026Masahiko SawadamasterunreleasedREL_19_STABLE19rc1
LLM analysis, 100% surelanded as postedfrom patchset #394 of 4 reviewers saw the final version

All hunks in the four changed files match #39 after normalizing rebase offsets and index metadata, and the commit message matches too. The attachment and commit credit Bharath Rupireddy although Masahiko Sawada sent the source patchset.

  • Manusaw the final versiontrailerManu compared polling and latch-based versions, tested failure controls and stability in #23, measured latch overhead in #25 and #28, and tested #29 repeatedly in #30, after which the production fix changed only editorially.
  • Zsolt Parragisaw the final versiontrailerZsolt endorsed and extended #20 in #22, defended its simpler design in #26, and carried the final production logic forward unchanged while revising tests in #35.
  • Daniel Gustafssonsaw the final versiontrailerDaniel verified the reproducer and check-world with #20 in #21, then endorsed the production patch and reviewed tests from #32 in #33, with only light production changes afterward.
  • Masahiko Sawadasaw the final versiontrailerMasahiko tested the initial callback fix in #18, reviewed the latch design and comments in #24 and #31, revised and repeatedly tested #39, and committed the final version.

Aug 2026

1 commit
Propagate rebalanced cost limit to parallel vacuum workers
28 Aug 2026Daniel GustafssonmasterunreleasedREL_19_STABLE19beta4
LLM analysis, 97% suremodified before commitlight changesfrom patchset #143 of 3 reviewers saw the final version

The vacuum.c fix, autovacuum.c injection point, and 001_parallel_autovacuum.pl changes exactly match #14 modulo rebase metadata. Daniel edited the commit-message explanation and added the backpatch rationale; no code or test changes were made.

  • Bharath Rupireddysaw the final versiontrailerBharath accepted the original approach in #6 and reported successful tests and pgindent for #9 in #11, followed only by test-stability and editorial changes.
  • Masahiko Sawadasaw the final versiontrailerMasahiko identified the second-worker timing race in #8 and approved its injection-point fix in #13, after which only test thresholds and commit prose changed.
  • Daniel Gustafssonsaw the final versiontrailerDaniel reviewed the original fix and test risks in #2, endorsed the dedicated injection point in #10, took #14 through final validation, and confirmed committing it in #16.

Credits across 2 commits

Author
Bharath Rupireddy1
Zsolt Parragi1
Reviewer
Daniel Gustafsson2
Masahiko Sawada2
Bharath Rupireddy1
Manu1
Zsolt Parragi1
Reported by
Nikolay Samokhvalov1
Committer
Daniel Gustafsson1
Masahiko Sawada1