👋 , In our Flink deployments, often times there’s a use case where we would want the flink pipeline, that was previously running with state, to start from scratch, as in from no state. When using the Flink K8s operator, would it mean we can change a running production pipeline spec from

upgradeMode: savepoint

to

upgradeMode: stateless

? Operator docs mention

stateless

isn’t recommended for production use, is that a general recommendation or that

stateless

wouldn’t work for this case ? Thanks !

I have this workflow implemented by deleting the current deployment and then creating a new one with exactly the same configuration, but no initial savepoint set. Works great 🙂 Curious to see if there is a simpler way to do it.

👋 Yaroslav ! The steps you mentioned seem to be similar to https://nightlies.apache.org/flink/flink-kubernetes-operator-docs-main/docs/custom-resource/job-management/#manual-recovery And, yes, while that can work, it is manual and slightly complex. Curious to know if there’s an easier way.

Stateless upgradeMode should start the job with empty state , but this is not thoroughly tested I think

Thanks folks ! We’d go with the delete and recreate for now