SEZ9 commented on PR #11077: URL: https://github.com/apache/seatunnel/pull/11077#issuecomment-5366240306
Thanks @DanielLeens for the re-verification on head `27da28820`. On the two blockers, I agree they remain unchanged on this head: 1. **Issue 1 (duplicated lifecycle calls)** — since the same writer instance is registered under N `SinkIdentifier` keys, `snapshotState`/`prepareCommit`/`abortPrepare`/`close` fire N times per checkpoint on one writer. This needs identity-based dedup (iterating a distinct set of writer instances) before those lifecycle calls. 2. **Issue 2 (state lookup inside `computeIfAbsent`)** — only the first aliased table's state is ever read on restore; the lookup needs to be hoisted out so state for all `SinkIdentifier`s mapping to the same destination is collected and passed to `restoreWriter`. On the newly reported compile failure and the proposed fix commit `f2a9c4ddb31ffd849faa12f13f0afb2a11baaacb`: the 2026-08-19 full re-review of the same head `27da28820` didn't surface a compilation failure, so before we act on that patch, could you share a link to the failing CI run or build log showing the error? Once we can confirm the failure, we can evaluate the fix on its merits. The writer-creation dedup itself remains the right approach. Once Issues 1 and 2 are addressed I'm happy to do another full pass. Thanks for keeping this moving. <!-- streview-comment:413 --> -- 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]
