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

Posted to dev@flink.apache.org by "Gyula Fora (Jira)" <ji...@apache.org> on 2022/12/16 13:45:00 UTC

[jira] [Created] (FLINK-30437) Last-state upgrade combined with state incompatibility causes state loss

Gyula Fora created FLINK-30437:
----------------------------------

             Summary: Last-state upgrade combined with state incompatibility causes state loss
                 Key: FLINK-30437
                 URL: https://issues.apache.org/jira/browse/FLINK-30437
             Project: Flink
          Issue Type: Bug
          Components: Kubernetes Operator
    Affects Versions: kubernetes-operator-1.3.0, kubernetes-operator-1.2.0
            Reporter: Gyula Fora
            Assignee: Gyula Fora


Even though we set:
execution.shutdown-on-application-finish: false
execution.submit-failed-job-on-application-error: true
If there is a state incompatibility the jobmanager marks the Job failed, cleans up HA metada and restarts itself. This is a very concerning behaviour, but we have to fix this on the operator side to at least guarantee no state loss.

The solution is to harden the HA metadata check properly (like we tried but failed in the past :) )



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