SEZ9 commented on PR #12173: URL: https://github.com/apache/seatunnel/pull/12173#issuecomment-6282015644
@Rangsh thanks for the quick turnaround and for keeping it a Javadoc-only commit on top of `7948e2640` — no logic change and no rebase is exactly what makes re-verification cheap. On your ask: agreed, I will not carry the earlier verification over. I'll do a fresh F1–F6 pass against `b137173f7` rather than against `c0d59da`. For F1 specifically, the three statements you listed are what I was after: which iterations are sampled (first `initialStoredJobCount=0` iteration plus trial tear-down, everything else resident-only), what the unsampled window can and cannot hide (a missing resident entry is still caught; a write-through WAL append that starts failing after the first sample surfaces only at tear-down), and the pointer to #12266 for widening the tear-down sample from one representative key to the whole batch. That also covers the tracking link I wanted for F3. If the Javadoc reads as you've summarised it here, F1 and F3 close on `b137173f7`. Two small things to settle while I go through the rest: 1. **F4** — you describe `b137173f7` as Javadoc-only. Is it intentional that `docs/en|zh/engines/zeta/benchmark.md` does not get even a one-line note on the sampling window for `runningJobGrowth` / `completedJobHistoryGrowth`? I'm fine with "Javadoc is the source of truth" if that's the call, but please say so explicitly so the record is clear. 2. **F2 / F5 / F6** — I'm not aware of any changes to `JobHistoryService` or `JobHistoryServiceFinishedMetricsTest` since `ca82f95d3b` / `909f878402` / `c0d59da54`. If `b137173f7` touched them at all, let me know; otherwise I'll re-confirm those files as unchanged on the new head. Same-runner `Benchmarks` and the ~19% CV at count 1000 stay observational and non-blocking, as discussed. Once you answer the F4 question I'll post the F1–F6 confirmation against `b137173f7`. <!-- streview-comment:1886 --> -- 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]
