Jul 2021

4 commits
Close yet another race condition in replication slot test code
29 Jul 2021Alvaro Herreramaster15beta1REL_19_STABLE15beta1REL_18_STABLE15beta1REL_17_STABLE15beta1REL_16_STABLE15beta1REL_15_STABLE15beta1REL_14_STABLE14beta3REL_13_STABLE13.4
LLM analysis, 100% suremodified before commitlight changesfrom patchset #241 of 2 reviewers saw the final version

Preserves the explicit checkpoints and temporary max_wal_size setting from #24, with comment and formatting edits in src/test/recovery/t/019_replslot_limit.pl. It also corrects an outdated xlog.c comment, as confirmed in #25.

  • Tom Lanesaw an earlier versionLLM analysis, 94% sureReported continuing buildfarm failures in the committed #18 test at #21, prompting the subsequent change to explicit checkpoint scheduling.
  • Alvaro Herrerasaw the final versioncommittercommitted the patch
Make new replication slot test code even less racy
19 Jul 2021Alvaro Herreramaster15beta1REL_19_STABLE15beta1REL_18_STABLE15beta1REL_17_STABLE15beta1REL_16_STABLE15beta1REL_15_STABLE15beta1REL_14_STABLE14beta3REL_13_STABLE13.4
LLM analysis, 100% suremodified before commitlight changesfrom patchset #181 of 1 reviewer saw the final version

The checkpoint-completion polling loop and test-count change match #18; only its introductory comment and assertion description differ in src/test/recovery/t/019_replslot_limit.pl.

  • Alvaro Herrerasaw the final versioncommittercommitted the patch
Make new replication slot test code less racy
17 Jul 2021Alvaro Herreramaster15beta1REL_19_STABLE15beta1REL_18_STABLE15beta1REL_17_STABLE15beta1REL_16_STABLE15beta1REL_15_STABLE15beta1REL_14_STABLE14beta3REL_13_STABLE13.4
LLM analysis, 100% suremodified before commitlight changesfrom patchset #17

The warning-polling replacement matches #17 exactly except that the assertion description says slot invalidation has been logged rather than that the slot has been invalidated.

Advance old-segment horizon properly after slot invalidation
16 Jul 2021Alvaro Herreramaster15beta1REL_19_STABLE15beta1REL_18_STABLE15beta1REL_17_STABLE15beta1REL_16_STABLE15beta1REL_15_STABLE15beta1REL_14_STABLE14beta3REL_13_STABLE13.4
LLM analysis, 100% suremodified before commitlight changesfrom patchset #131 of 2 reviewers saw the final version

Matches the executable changes in #13 across xlog.c, slot.c, slot.h, and the recovery test. Committer edits rewrite checkpoint comments, document KeepLogSeg's invalidation requirements, and tidy test formatting.

  • Jeff Janessaw an earlier versionLLM analysis, 88% sureReported successful testing and suggested clearer wording in #9, quoting the comment from #5, before later changes to invalidation handling and resource-limit recalculation.
  • Alvaro Herrerasaw the final versioncommittercommitted the patch

Credits across 4 commits

Author
Kyotaro Horiguchi3
Alvaro Herrera1
Reviewer
Alvaro Herreracommitter3
Reported by
Marcin Krupowicz1
Committer
Alvaro Herrera4