[
https://issues.apache.org/jira/browse/FLINK-29258?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17603059#comment-17603059
]
Matyas Orhidi commented on FLINK-29258:
---------------------------------------
The configuration
h5. kubernetes.operator.job.upgrade.last-state-fallback.enabled
does exactly this.
h5.
> Add SAVEPOINT_ONLY upgrade mode
> -------------------------------
>
> Key: FLINK-29258
> URL: https://issues.apache.org/jira/browse/FLINK-29258
> Project: Flink
> Issue Type: New Feature
> Components: Deployment / Kubernetes
> Reporter: Maximilian Michels
> Priority: Major
>
> Users may want to explicitly upgrade from a savepoint. the {{SAVEPOINT}}
> upgrade mode cannot ensure this. When the savepoint creation fails, which
> might be temporary, the operator falls back to using the latest checkpoint.
> To prevent this from happening, we could add a SAVEPOINT_ONLY upgrade mode
> which does not proceed when the savepoint creation fails.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
