SbloodyS commented on issue #18698: URL: https://github.com/apache/dolphinscheduler/issues/18698#issuecomment-6142472322
Based on the 3.4.3 source, this report does not yet establish a separate crash/failover bug. Although persisting the task failure and advancing the in-memory graph are separate operations, recovery includes an event replay path: WorkflowFailoverCommandHandler rebuilds the graph using existing task instances, and TaskFailureStateAction.onStartEvent() republishes TaskFailedLifecycleEvent for a task already in FAILURE. That event re-enters the failure-chain marking and topology transition logic. Additionally, workflowFinish() updates the workflow state and deletes its serial command within the same transaction. A FAILURE task belonging to a RUNNING_EXECUTION workflow is also insufficient evidence by itself: the task may be awaiting retry, or other branches may still be executing under the Continue failure strategy. The absence of an independent serial-command reconciliation mechanism is a potential resilience improvement, but it does not demonstrate that the existing failover recovery path fails. To confirm this as a bug, please provide: - The exact affected version/build, preferably with reproduction on 3.4.3. - A minimal workflow export showing the DAG, failure strategy, task retry settings, and serial execution mode. - The precise crash injection point and the Master topology, including how restart/failover was performed. - Master logs covering the failure, failover command processing, recovered workflow startup, and subsequent task/workflow lifecycle events. - Timestamped task, workflow, and serial-command records before the crash and after recovery, showing that the workflow remains stuck after retries and other branches have finished. - Evidence that a later serial instance remains blocked, including its command/instance state. The key missing evidence is where the recovered execution stops progressing despite the failure-event replay path. That would help distinguish a recovery defect from the previously reported DAG issue or an expected intermediate state. -- 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]
