Rangsh commented on issue #12063: URL: https://github.com/apache/seatunnel/issues/12063#issuecomment-5595572471
Thanks @SEZ9 — replied on the PR review and pushed `1de359ef9` on https://github.com/apache/seatunnel/pull/12173. ### Your two open questions 1. **`completedJobHistoryGrowth` + `initialStoredJobCount=1000` (+ GC)** Same trial-oriented / light-sampling durability shift applies to both growth methods. Dispatching same-runner fork Diagnostics for `completedJobHistoryGrowth` @ `1000` on the PR head now: https://github.com/Rangsh/seatunnel/actions/runs/34309225279 Next: matching `dev` before run (queued one-at-a-time), then I will post wall/GC/JFR frame counts the same way as for `runningJobGrowth` @0. 2. **`runningJobGrowth` @0 CV local regression (20.9% → 31.7%)** That was a **local** mixed run, not the same-runner Diagnostics pair. On the fork runner, the before/after wall+GC+JFR for `runningJobGrowth` @0 showed the intended structural win (`reloadFromMapStore` 8→3, `WALReader.loadAllData` 5→1), so I treat the local CV bump as machine noise unless the same-runner JMH scores disagree. Happy to re-run that pair if you want another sample after `1de359ef9` (sampled durability is slightly denser than trial-only). ### PR review items (Issues 1–6) Addressed in `1de359ef9`: light per-sample durability + suppressed cleanup, IMap key-lock around metrics merge, full-batch reload, en/zh benchmark docs, contract-only Javadoc, and stronger unit tests. Details on the PR comment. -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected]
