surafel58 commented on PR #11778: URL: https://github.com/apache/seatunnel/pull/11778#issuecomment-5278293251
Thanks for the deep review. Pushed 9608d8a1 with the fixes: - Issue 1 (Spark/Flink): documented the limit in incompatible-changes.md and the Timer Flush section. sink.flush.interval is Zeta-only; on Spark and Flink there is no periodic flush, so the buffer is sent by batch_size and on close. I kept flush_interval removed (per the earlier review and the other FlushSignal-migrated connectors) and documented the capability loss instead of re-adding a connector thread. - Issue 2: corrected the validation wording. A leftover flush_interval key only fails under --check/--dry-run; a direct run ignores it. Added a one-time WARN in the writer when the raw config still has flush_interval. - Issue 3: removed the 'on checkpoint' claim from the Timer Flush docs (EN/ZH), since PrometheusWriter does not override prepareCommit(). - Issue 6: close() now closes the writer's own HttpClientProvider (it shadowed the parent field and leaked). - Issue 7: added a comment explaining the context null-check. - 2.2: added a unit test that the buffer is still flushed on close when the engine never invokes the flush action (Spark/Flink path). The red Build is the flaky windows engine test (RestApiSubmitJobConfigShadeDecryptTest / Node failed to start), unrelated to this change. -- 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]
