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 2022/09/21 08:08:00 UTC

[jira] [Created] (FLINK-29376) Deployment already exists error if Flink version is not set correctly

Gyula Fora created FLINK-29376:
----------------------------------

             Summary: Deployment already exists error if Flink version is not set correctly
                 Key: FLINK-29376
                 URL: https://issues.apache.org/jira/browse/FLINK-29376
             Project: Flink
          Issue Type: Bug
          Components: Kubernetes Operator
    Affects Versions: kubernetes-operator-1.1.0, kubernetes-operator-1.2.0
            Reporter: Gyula Fora


If the user incorrectly sets a Flink version lower than 1.15 when using 1.15 or above there are some strange behaviour currently around cluster shutdown.

Since we always set SHUTDOWN_ON_APPLICATION_FINISH (in the FlinkConfigBuilder) regardless of version but set the JmDeployStatus to MISSING/READY based on version. It can happen that we set the JmStatus to MISSING when the jm deployment is still running.

As the observer skips observing UPGRADING CRs it never updates the MISSING status and the deployment logic fails on duplicate deployment.

An easy fix could be to only set SHUTDOWN_ON_APPLICATION_FINISH for version >= 1.15



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