SEZ9 commented on issue #13298: URL: https://github.com/apache/seatunnel/issues/13298#issuecomment-7025827574
Thanks @1328837476-hug, that plan matches the staged approach: baseline first, design note once #13060 has established its final recovery contract, and no code PR until the design has maintainer agreement. The issue is still formally unassigned, but that does not block you from preparing the design evidence. Remaining asks for when you post the note: - **Measurements:** share the producer-initialization counts and recovery times together with the writer count, retained transaction history, and retry count for each scenario. Please also state which revision the initial baseline and the rerun were each run against, so the two sets are comparable. - **Coverage and ownership:** show how every legacy prefix is identified from persisted, engine-neutral state, and how the per-prefix cleanup owner is determined. - **Owner failure:** describe how a replacement safely resumes cleanup after the owner fails or retries. - **Redistribution:** cover mixed prefixes and the actual Flink and Zeta state-redistribution cases. - **Verification:** include the `read_committed` recovery matrix. Until that design is agreed, the current per-writer cleanup behaviour stays as is, and this work should not change #13060. <!-- streview-comment:2475 --> -- 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]
