[
https://issues.apache.org/jira/browse/FLINK-27095?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17518165#comment-17518165
]
Gyula Fora commented on FLINK-27095:
------------------------------------
The Flink cluster itself will have to be deleted anyways, so having to delete
the FlinkDeployment is a good signal to the user to show what is happening
underneath.
The only way to create the wrong one by accident is to define/not define job
properties. I think deleting the deployment and recreating it in these very
rare cases is not a problem at all. It's a trivial operation.
So I would say this works exactly as expected.
> Allow FlinkDeployment Mode Switching
> ------------------------------------
>
> Key: FLINK-27095
> URL: https://issues.apache.org/jira/browse/FLINK-27095
> Project: Flink
> Issue Type: Improvement
> Components: Kubernetes Operator
> Reporter: Usamah Jassat
> Priority: Major
> Fix For: kubernetes-operator-1.0.0
>
>
> Currently the operator doesn't allow switching the mode of an existing
> FlinkDeployment between Application and Session.
> I don't see why this shouldn't be allowed, if the mode is switched we should
> tear-down the cluster and spin-up a new cluster running in this new mode. It
> will add complexity into the connector but seems like a valid use-case.
--
This message was sent by Atlassian Jira
(v8.20.1#820001)
