Oct 2023

2 commits
Try to handle torn reads of pg_control in frontend.
16 Oct 2023Thomas Munromaster17beta1REL_19_STABLE17beta1REL_18_STABLE17beta1REL_17_STABLE17beta1REL_16_STABLE16.1REL_15_STABLE15.5REL_14_STABLE14.10REL_13_STABLE13.13+1 more
LLM analysis, 100% surepart of a posted patchsetlight changesfrom patchset #292 of 3 reviewers saw the final version

Extracts the frontend-retry component of #29, preserving its ten-retry limit, repeated-bad-CRC stopping condition and 10 ms sleep, with only the explanatory comment rewritten. Git ancestry identifies this as the development-branch commit and 5725e4ebe7 as its identical REL_16_STABLE counterpart, consolidated here under the one-verdict-per-change rule.

  • Anton A. Melnikovsaw the final versiontrailerAnton tested the earlier retry and locking implementations in #9, #12, #18 and #20, then reviewed the final retry component and proposed moving its loop to callers in #31.
  • David Steelesaw the final versiontrailerDavid reported that the series looked generally reasonable in #30 and explicitly supported committing the frontend retries in #35.
  • Michael Paquierno review foundtrailerMichael diagnosed a possible race in #2 and reviewed other components in #34, but the supplied discussion contains no identifiable review of the frontend-retry patch.
Acquire ControlFileLock in relevant SQL functions.
15 Oct 2023Thomas Munromaster17beta1REL_19_STABLE17beta1REL_18_STABLE17beta1REL_17_STABLE17beta1REL_16_STABLE16.1REL_15_STABLE15.5REL_14_STABLE14.10REL_13_STABLE13.13+2 more
LLM analysis, 100% surepart of a posted patchsetlight changesfrom patchset #292 of 3 reviewers saw the final version

Extracts the SQL-function component of #29 unchanged: src/backend/utils/misc/pg_controldata.c gains the lwlock include and shared ControlFileLock acquisition/release around four get_controlfile() calls. The other three components are absent.

  • David Steelesaw the final versiontrailerDavid reviewed the series in #30 and explicitly agreed to committing its SQL-locking component in #35.
  • Anton A. Melnikovsaw an earlier versiontrailerAnton tested and reviewed earlier solutions for SQL control-file reads in #9, #12, #18 and #20, before filesystem locking was replaced with explicit SQL-function LWLocks.
  • Michael Paquiersaw the final versiontrailerMichael explicitly accepted the SQL-locking component, including its suitability for PostgreSQL 11, in #34.

Credits across 2 commits

Author
Thomas MunroLLM2
Reviewer
Anton A. Melnikov2
David Steele2
Michael Paquier2
Committer
Thomas Munro2

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