nzw921rx commented on issue #12063: URL: https://github.com/apache/seatunnel/issues/12063#issuecomment-5601271546
> ### Same-runner evidence: `completedJobHistoryGrowth` @ `initialStoredJobCount=1000` > Fork Diagnostics (`java_version=11`, `profile=wall`, `capture_jfr=true`, `jmh_args=-p initialStoredJobCount=1000 -wi 1 -i 3`): > > run wall summary > **before** (`dev`) https://github.com/Rangsh/seatunnel/actions/runs/34319489970 `WALReader.loadAllData` **30** > **after** (`5bbc304b2`) https://github.com/Rangsh/seatunnel/actions/runs/34318323026 `WALReader.loadAllData` **15** > Same structural direction as `runningJobGrowth` [@0](https://github.com/0): fewer full-WAL reloads under the growth fixture. PR head also keeps whole-batch **resident** checks every iteration, mid-trial MapStore reload only when `initialStoredJobCount=0` (full-batch `loadAll` under pressure=1000 OOMs the diagnostic heap — see PR notes), and trial durability via one representative key + whole-batch resident assert. > > Local smoke on the after head (`-Xmx2g -f1 -wi1 -i2`): `completedJobHistoryGrowth` [@1000](https://github.com/1000) ≈ **2033 us/op**, no OOM. thanks,It is necessary to use a normal benchmark workflow, as the flow of diagnostic work will be affected by profile monitoring and cannot be compared between them -- 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]
