[
https://issues.apache.org/jira/browse/FLINK-31794?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17713908#comment-17713908
]
Gyula Fora commented on FLINK-31794:
------------------------------------
But even then I don't understand how any change in the operator would solve the
problem because the Kubernetes API will simply reject objects with invalid
schemas.
I think we need to introduce the `additionalProperties` fields in the CRD for
the different objects
> Ignore unknown CRD fields
> -------------------------
>
> Key: FLINK-31794
> URL: https://issues.apache.org/jira/browse/FLINK-31794
> Project: Flink
> Issue Type: Improvement
> Components: Kubernetes Operator
> Affects Versions: kubernetes-operator-1.4.0
> Reporter: Daren Wong
> Priority: Minor
> Labels: pull-request-available
> Fix For: kubernetes-operator-1.4.0
>
>
> Currently, Kubernetes Operator fails to create a CR (FlinkDeployment or
> FlinkSessionJob) if there is a new field in the CR that's not included in the
> CRD.
> This change is to propose removing this restriction for
> FlinkSessionJobStatus, FlinkSessionJobSpec, FlinkDeploymentStatus,
> FlinkDeploymentSpec by adding class-level annotation
> `{color:#bbb529}@JsonIgnoreProperties{color}(ignoreUnknown =
> {color:#cc7832}true{color})`.
>
> This means that any new fields in CR will be ignored if it's not registered
> in CRD preventing failures to create the CR. This helps users who need to
> deploy CRD and Kubernetes Operator separately to ensure backward
> compatibility.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
