You are viewing a plain text version of this content. The canonical link for it is here.

Posted to issues@flink.apache.org by "Sylwester Lachiewicz (Jira)" <ji...@apache.org> on 2026/07/13 21:21:00 UTC

[jira] [Created] (FLINK-40137) Extract ReaderRecoveryCoordinator from DynamicKafkaSourceEnumerator

Sylwester Lachiewicz created FLINK-40137:
--------------------------------------------

             Summary: Extract ReaderRecoveryCoordinator from DynamicKafkaSourceEnumerator
                 Key: FLINK-40137
                 URL: https://issues.apache.org/jira/browse/FLINK-40137
             Project: Flink
          Issue Type: Improvement
            Reporter: Sylwester Lachiewicz


▎ The enumerator accumulated three interacting pieces of recovery-gating state (initial registration pending, reported splits pending redistribution, deferred metadata-update readers) inline, making the gating semantics hard to follow and untestable without a full container-based test setup. Extract this state and its logic into a dedicated ReaderRecoveryCoordinator class with its own unit tests. No behavior change.



--
This message was sent by Atlassian Jira
(v8.20.10#820010)