[
https://issues.apache.org/jira/browse/FLINK-30527?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Gyula Fora closed FLINK-30527.
------------------------------
Fix Version/s: kubernetes-operator-1.4.0
kubernetes-operator-1.3.1
Resolution: Fixed
main 0dd9e9c257b1e4d7659d8c2cb367a35e7736c7a1
release-1.3 7f65de716cb7a2d6ac158cd01f1ee2923b634795
> Last-state suspend followed by flinkVersion change may lead to state loss
> -------------------------------------------------------------------------
>
> Key: FLINK-30527
> URL: https://issues.apache.org/jira/browse/FLINK-30527
> Project: Flink
> Issue Type: Bug
> Components: Kubernetes Operator
> Affects Versions: kubernetes-operator-1.2.0, kubernetes-operator-1.3.0
> Reporter: Gyula Fora
> Assignee: Gyula Fora
> Priority: Critical
> Labels: pull-request-available
> Fix For: kubernetes-operator-1.4.0, kubernetes-operator-1.3.1
>
>
> We do not check flink version changes on recovery, so if the user suspended
> the job using last-state mode (or that was the only mode available) and then
> subsequently the flinkVersion is changed to a non-HA compatible version, the
> job would be restored using last-state and state would be lost.
> In these cases we should set an error in the Flink resource instructing the
> user that changing version is not allowed at that point.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
