SEZ9 commented on PR #12048: URL: https://github.com/apache/seatunnel/pull/12048#issuecomment-5862682834
Thanks for the detailed re-review at `56794bc19d3e` (review `#5325863940`). Going through F1–F8: - **F1 / F5** — Documenting the recovery procedure and the least-privilege rationale for the `EntityPath` rejection is the direction I suggested. I haven't seen the docs diff itself in this thread, though, so before closing F1 (HIGH) could you point me at the actual paragraph so I can confirm it states that the job must be restarted *without* restoring the old checkpoint state? - **F2** — If the connector's shade execution inherits the parent `ServicesResourceTransformer` with no `<transformers>` override, that should resolve it. Could you confirm the relocated Azure/Netty/Proton service provider files are actually present in the packaged connector jar? - **F3 / F6** — The described test coverage and shared builder sound right. I'd like to verify against the diff rather than the description before marking them resolved — a pointer to the relevant hunks would be enough. - **F4** — Agreed it is non-blocking. Remaining ask: an explicit tracking reference (follow-up ticket or issue) for the shaded Netty/Jackson/Proton versions rather than a one-time disclosure. - **F7 / F8** — Your comment appears to be cut off partway through F7, so I can't confirm either from the thread. Could you restate (a) the behaviour when the live partition count exceeds the checkpointed set (fail-fast vs. log-and-continue), and (b) what, if anything, was done for connector-level reconnect/retry tuning, or where the SDK-default retry window is documented for operators? Once those are in, I'm happy to move this forward. <!-- streview-comment:1375 --> -- 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]
