[
https://issues.apache.org/jira/browse/FLINK-36893?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17905214#comment-17905214
]
Ryan van Huuksloot commented on FLINK-36893:
--------------------------------------------
We use the Kubernetes Operator which has this functionality
|h5. kubernetes.operator.periodic.checkpoint.interval|
> Support periodic full checkpoint
> --------------------------------
>
> Key: FLINK-36893
> URL: https://issues.apache.org/jira/browse/FLINK-36893
> Project: Flink
> Issue Type: Improvement
> Components: Runtime / Checkpointing
> Reporter: Xingcan Cui
> Priority: Minor
>
> We have a few Flink jobs running with incremental checkpoints. The S3 bucket
> used to store the checkpoint files has an s3.entropy.key set and is
> automatically cleaned up with a time-based life cycle rule. However, the rule
> occasionally removes some pretty old checkpoint files that are still
> referenced by the latest checkpoint and that will make the job unrecoverable.
> It would be good if Flink can provide a periodic full checkpoint option that
> allows the checkpoint files to be fully refreshed.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
