Hello, I was wondering what the expected checkpoint behavior is for a task manager that receives a sigterm. Will it allow any time to complete an in progress async checkpoint to cloud storage? If so, is there any configuration to control how long it waits? This is for a flink streaming application that is processing data from kafka

Not sure if there is a good approach to this. You typically want to use savepoints to stop and restarts the jobs smoothly.

I managed to find a way using a k8s prestop hook since we’re using the k8s operator. For context this is for more smooth handling of node rotations on the underlying k8s cluster, which can happen without prior notice in this case