Rangsh commented on issue #12058: URL: https://github.com/apache/seatunnel/issues/12058#issuecomment-5692860645
@SEZ9 thanks — addressing the three remaining asks. **1. Scope note in the gist report** — done. `NOISY-ALIGN-REPORT.md` now opens with an explicit one-line scope note that all conclusions apply only to the `file:///` LocalFileSystem harness on `38a37104e` and must not be generalized to a real HDFS/DFS client. Updated gist: https://gist.github.com/Rangsh/3c3f49c79c6583a1ca954b7d47a49627 **2. Diagnostic system property removed** — confirmed. The OPI=1 TearDown skip (`-Dseatunnel.benchmark.diag.skipOverviewValidate=true`) and the temporary `CHECKPOINT_OPERATIONS_PER_INVOCATION = 1` patch were worktree-only and were restored via the run script’s `trap`. Post-run check: both the main tree and the wait worktree have `CHECKPOINT_OPERATIONS_PER_INVOCATION = 100`, durable reload/validate TearDown restored, and `seatunnel.benchmark.diag.skipOverviewValidate` is absent from the fixture/workload sources. **3. Production / issue status** — unchanged. No production change from this evidence; `#12081` stays Related-only and does not close `#12058`. If the historical high-CV day (~32% / ~393 µs) reappears, I will capture it with the same baseline, JDK, JMH defaults, `file:///` store, and 0 ns `ThreadPark` JFR settings used for attempt-04 so the runs are directly comparable. -- 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]
