DanielLeens commented on issue #11777: URL: https://github.com/apache/seatunnel/issues/11777#issuecomment-5266691913
Thanks for writing this STIP down with the runtime semantics separated from dirty-data collection. I re-checked the current `dev` engine path before replying here, and the problem statement is grounded: member removal still flows through `SeaTunnelServer.memberRemoved(...)` -> `CoordinatorService.memberRemoved(...)` -> `failedTaskOnMemberRemoved(...)`, pipeline restore counting is still pipeline-local through `SubPlan.pipelineRestoreNum`, and the current public job-result path still does not expose a durable job-level "recovery degraded" contract. For v1, I would recommend keeping the public boundary narrow and additive: 1. do not change `JobStatus` or retry semantics; 2. make the dirty/recovery summary an additive read model with disabled-by-default behavior; 3. treat incomplete tracking as `UNKNOWN`, not `CLEAN`; 4. keep planned-leave exclusion out of scope until there is a trusted control-plane signal, not a self-reported shutdown hook; 5. keep the review centered on episode/attempt deduplication, active-master recovery, and bounded finished-history persistence. If you keep Phase 1 within that boundary, this looks like a reasonable STIP to continue with a draft PR. -- 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]
