Skip to content

[BugFix] Update parallelgroup cache when entrance instance goes offline (#5407) - #5478

Open
zhang-arvin wants to merge 1 commit into
apache:masterfrom
zhang-arvin:fix/5407-parallelgroup-cache-offline
Open

zhang-arvin wants to merge 1 commit into
apache:masterfrom
zhang-arvin:fix/5407-parallelgroup-cache-offline

Conversation

@zhang-arvin

Copy link
Copy Markdown

What changes were proposed in this pull request?

When an entrance instance goes offline (via /markoffline) or comes back online, the cached per-user parallel groups are not refreshed, so jobs keep being routed with concurrency recalculated from a stale entrance count.

This PR wires cache refresh into the offline/online transitions:

  • EntranceGroupFactory now records each cached group's creator/user/engineType key and exposes refreshAllGroups(), which re-queries the user configuration and recomputes maxRunningJobs with the latest getRunningEntranceNumber().
  • EntranceUserParallelOrchestratorPlugin exposes invalidateCache() to drop the 30-minute configCache entries.
  • EntranceLabelRestfulApi (/markoffline and /backonline) triggers both refreshes after the RouteLabel update.
  • EntranceOrchestrationFactory gets a invalidateUserParallelPluginCache() entry point (Scala 2.11-safe companion class for Java callers).

Closes #5407

@zhang-arvin

Copy link
Copy Markdown
Author

Base-branch note: the failing checks are not caused by this PR.

All of sql-check, spotless-check, third-party-dependencies-check and build-backend fail on files that this PR does not touch, and they fail identically on the other open PRs from the same branch point. The failures come from master itself:

None of these files appears in this PR's diff, so there is nothing on this PR to fix here; it will go green once master is repaired. This PR stays on the master baseline problem until that lands.

This comment was generated by an AI agent (Hermes Agent).

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

[bug][CGS][entrance] parallelgroup cache not updated when entrance instance goes offline

1 participant