fxbing opened a new pull request, #4573: URL: https://github.com/apache/flink-cdc/pull/4573
## What is the purpose of this pull request? Fix [FLINK-40973](https://issues.apache.org/jira/browse/FLINK-40973). Periodic Fluss source discovery currently triggers job failover on transient metadata errors. This change preserves the last successful discovery state and lets the next scheduled cycle retry when the exception chain contains a Fluss `RetriableException`. ## Brief change log - Log a warning and skip the failed discovery result only when periodic discovery is enabled and the error is retriable. - Preserve subscriptions, assigned and pending splits, and pending-removal state. Non-retriable errors and one-time discovery still fail. - Add regressions for initial discovery, state preservation, subsequent recovery, and errors that must still fail. ## Verifying this change - Three regression instances reproduce the original discovery exception before the fix and pass afterward. - Java 11 / Flink 1.20.3: 9 focused Enumerator tests passed. - Java 17 / Flink 2.2.0: 6 focused Enumerator tests passed. - Both runs completed without failures, errors, or skipped tests. Spotless and Checkstyle passed. The regressions inject a wrapped `NetworkException` through `TableDiscoverer`. A full pipeline leader-failover end-to-end test was not run. ## Documentation - Does this pull request introduce a new feature? No. - If yes, how is the feature documented? Not applicable. --- ##### Was generative AI tooling used to co-author this PR? - [X] Yes — Codex CLI -- 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]
