Hi all, I am trying to minimize the impact of random container restarts on the end to end latency of my flink application running on k8s. I have a stateful app that uses kafka for both source & sink. When a container receives SIGTERM, I imagine flink has some housekeeping stuff to do in order to shut down gracefully. Is there a hook that I can plug into to make sure all the TM's completed whatever they needed to do during shutdown gracefully? Am I even thinking about this correctly? aka would it be better to just tune this by kafka consumer/producer properties? btw, apps are deployed via apache-flink-operator 1.5

you might want to look into this: https://nightlies.apache.org/flink/flink-docs-master/docs/dev/datastream/fault-tolerance/checkpointing/