Feb 2023

3 commits
Un-revert "Disable STARTUP_PROGRESS_TIMEOUT in standby mode."
10 Feb 2023Robert HaasREL_15_STABLE15.3
LLM analysis, 100% surelanded as postedfrom patchset #1352 of 5 reviewers saw the final version

The reapplication has the same stable patch ID as #135, restoring its changes to xlogrecovery.c, startup.c and startup.h unchanged on PostgreSQL 15. Release timing was cleared in #142, and restoration was confirmed in #143.

  • Simon Riggssaw an earlier versionLLM analysis, 97% sureReviewed the earlier disabling approach and its helper functions in #118, #121 and #124, before disabling was centralized in EnableStandbyMode().
  • Thomas Munrosaw an earlier versionLLM analysis, 99% sureReported the unnecessary wakeups in #116 and approved the alternative implementation in #127, before the final transition-based design.
  • Kyotaro Horiguchisaw an earlier versionLLM analysis, 90% sureRaised robustness concerns about coupling timer disabling to standby-mode assignments in #131, before the subsequent guard changes and return to the simpler version.
  • Tom Lanesaw the final versionLLM analysis, 96% sureExplicitly accepted reinstating the unchanged backport after the release in #142.
  • Robert Haassaw the final versioncommittercommitted the patch
Revert "Disable STARTUP_PROGRESS_TIMEOUT in standby mode."
6 Feb 2023Robert HaasREL_15_STABLE15.2reverts 1 commit here
LLM analysis, 100% surewritten from the discussion, no matching patch1 of 1 reviewer saw the final version

Exactly reverses the PostgreSQL 15 backport of the standby-timeout fix; no attached patch proposes this reversal. Tom requested postponement because of release-wrap timing in #139, and Robert confirmed the revert in #140.

  • Tom Lanesaw the final versionLLM analysis, 99% sureIdentified the release-timing risk in #137 and requested reverting the backport in #139.
Disable STARTUP_PROGRESS_TIMEOUT in standby mode.
6 Feb 2023Robert Haasmaster16beta1REL_19_STABLE16beta1REL_18_STABLE16beta1REL_17_STABLE16beta1REL_16_STABLE16beta1REL_15_STABLE15.2
LLM analysis, 100% surelanded as postedfrom patchset #1351 of 4 reviewers saw the final version

The complete three-file diff has the same stable patch ID as #135, including EnableStandbyMode() and the enable/disable startup-timeout helpers. Robert confirmed commitment and backpatching in #136.

  • Simon Riggssaw an earlier versionLLM analysis, 99% sureDiscussed when disabling should occur and reviewed the helper-based implementation through #124, before the transition-site restructuring; the commit also credits him.
  • Thomas Munrosaw an earlier versionLLM analysis, 99% sureReported the wakeups in #116 and approved Robert's alternative in #127, before the final design; the commit also credits him.
  • Kyotaro Horiguchisaw an earlier versionLLM analysis, 90% sureReviewed the standby-mode coupling in #131 and preferred the reactive alternative for robustness, before later revisions.
  • Robert Haassaw the final versioncommittercommitted the patch

Oct 2021

3 commits
Fix race condition in startup progress reporting.
29 Oct 2021Robert Haasmaster15beta1REL_19_STABLE15beta1REL_18_STABLE15beta1REL_17_STABLE15beta1REL_16_STABLE15beta1REL_15_STABLE15beta1
LLM analysis, 100% surewritten from the discussion, no matching patch1 of 1 reviewer saw the final version

Matches Robert's two-line inline patch in #111: disable STARTUP_PROGRESS_TIMEOUT and clear startup_progress_timer_expired before beginning a new phase. No attachment contains this standalone fix, and commitment is confirmed in #115.

  • Nitin Jadhavsaw the final versionLLM analysis, 100% sureAnalyzed the stale flag in #112 and approved the complete race-prevention sequence in #114, matching the commit's review credit.
Report progress of startup operations that take a long time.
25 Oct 2021Robert Haasmaster15beta1REL_19_STABLE15beta1REL_18_STABLE15beta1REL_17_STABLE15beta1REL_16_STABLE15beta1REL_15_STABLE15beta1
LLM analysis, 99% surepart of a posted patchsetlight changesfrom patchset #1071 of 7 reviewers saw the final version

Commits only the startup-reporting component of the two-patch series in #107; that component has an identical stable patch ID. The repeating-timeout prerequisite landed separately as 732e6677a6.

  • Amul Sulsaw an earlier versionLLM analysis, 99% sureReported check-world failures and API issues in #41, then questioned implicit formatting arguments in #104, before repeating-timeout integration.
  • Bharath Rupireddysaw an earlier versionLLM analysis, 99% sureReviewed naming, redundant helpers and message structure in #43 and #45, then discussed single-user and bootstrap behavior in #54.
  • Justin Pryzbysaw an earlier versionLLM analysis, 99% sureSupplied callback corrections in #50, reported missing syncfs output and division by zero in #60, and tested standby and unit handling in #70 and #78.
  • Michael Paquiersaw an earlier versionLLM analysis, 99% sureIdentified the incorrect default time unit and discussed standby logging and configurability in #75, before later behavior and timer changes.
  • Álvaro Herrerasaw an earlier versionLLM analysis, 99% sureRequested clearer timeout API naming and translation extraction in #86, then reviewed scheduling semantics in #90 before the timer redesign.
  • Tom Lanesaw an earlier versionLLM analysis, 98% sureFlagged timestamp-polling overhead in #33 and proposed moving repeating-timeout state into timeout.c in #96, which drove the final restructuring.
  • Robert Haassaw the final versioncommittercommitted the patch
Add enable_timeout_every() to fire the same timeout repeatedly.
25 Oct 2021Robert Haasmaster15beta1REL_19_STABLE15beta1REL_18_STABLE15beta1REL_17_STABLE15beta1REL_16_STABLE15beta1REL_15_STABLE15beta1
LLM analysis, 100% surepart of a posted patchsetlight changesfrom patchset #1070 of 1 reviewer saw the final version

Commits the repeating-timeout component of #107 unchanged, with an identical stable patch ID; the startup-reporting component lands separately. Its Patch-by header credits Robert, and Nitin explicitly identifies it as Robert's unchanged submission from #106.

  • Tom Lanesaw an earlier versionLLM analysis, 97% sureProposed the reusable timeout facility in #96 and reviewed its clock-adjustment tradeoff in #101, before the explicit first-firing argument was added.

Credits across 6 commits

Author
Bharath RupireddyLLM2
Nitin JadhavLLM1
Robert HaasLLM1
Reviewer
Robert Haascommitter3
Committer
Robert Haas6

N LLM means N of the credits come from LLM analysis, not from commit trailers.