[
https://issues.apache.org/jira/browse/FLINK-38123?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=18009037#comment-18009037
]
Andrei Kaigorodov commented on FLINK-38123:
-------------------------------------------
[~gyfora] may I take this ticket, if so could you please assign it to me?
(reaching out to you as you reported at some point a similar ticket)
Apart from upgrading the base image to 1.20.1, does it make sense to
upgrade/move away from some deprecated APIs that the autoscaling example
currently uses which are:
- RuntimeContext.getNumberOfParallelSubtasks
- RuntimeContext.getIndexOfThisSubtask
- DiscardingSink
I can do it in the same PR in a separate commit (or can create a separate Jira
ticket if that is the way such things are handled)
> The autoscaling example image is based on Flink 1.18 rather than 1.20.1.
> ------------------------------------------------------------------------
>
> Key: FLINK-38123
> URL: https://issues.apache.org/jira/browse/FLINK-38123
> Project: Flink
> Issue Type: Bug
> Components: Examples, Kubernetes Operator
> Affects Versions: kubernetes-operator-1.12.1
> Reporter: Andrei Kaigorodov
> Priority: Minor
>
> since we build examples using flink 1.20.1 that is defined in the parent
> module:
>
> {code:java}
> <flink.version>1.20.1</flink.version>{code}
>
> we should align it with verion in
> flink-kubernetes-operator/examples/autoscaling/Dockerfile that is currently:
>
> {code:java}
> FROM flink:1.18{code}
> as a result, the example container doesn't start going to the crash loop.
>
> Additionally, we should use an image that uses java 17 i.e.
>
> {code:java}
> FROM flink:1.20.1-java17{code}
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
