SEZ9 commented on PR #12042: URL: https://github.com/apache/seatunnel/pull/12042#issuecomment-6460534720
Thanks for the follow-up on `2e03ae07114721f1312386cd75184a3e577f4782` — merging `dev` and resolving the conflict is appreciated. The re-review summary says F1, F2, F4, F7 and F8 are closed on this head, but I only have the prose and pseudo-code summary to go on, not the diff itself. Could you point me to the relevant hunks for each so I can confirm them on this head? For F7 in particular, please cover both halves: the `allMatch` short-circuit in `isCheckpointCoordinatorAlreadyStarted()` and the double iteration over the task list in the `PhysicalVertex` check. Still open from the earlier rounds (`73af28e` / `074283c`): - **F3 – state-only heuristic.** Both the `SubPlan` recoverability check and the new `restoreCoordinator(true)` logic derive "idle-closed" purely from `FINISHED` in `runningJobStateIMap`. Could you either cross-check against the restored checkpoint's own closed-task information, or add an explicit guard/log when a `FINISHED` unbounded-source group is not reflected in the restored checkpoint state? If you believe a state-only decision is safe, please spell out the reasoning in the PR description. - **F5 – coordinator vertices still strictly `RUNNING`.** A fully idle-closed source (enumerator also `FINISHED`) is still treated as "not started". If that is intentionally out of scope for this PR, please say so in the PR description and the `SubPlan` Javadoc; otherwise, let's extend the check. - **F6 – docs/release notes.** The PR declares a user-facing change, but the failover behavior change is not documented yet. A short note on how idle-closed readers are handled after a master switch would be enough. The re-review also mentions a test regression on this head, but I don't see the details. Could you share which test is affected and how you plan to fix it? Once the regression is resolved, F3/F5 are addressed or explicitly scoped out, and F6 has a doc entry, I'm happy to take another pass. <!-- streview-comment:2093 --> -- 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]
