Clemens Valiente
07/11/2024, 9:09 AM"RETAIN_ON_CANCELLATION": Checkpoint state is kept when the owning job is cancelled or fails.
but in the kubernetes implementation:
https://github.com/apache/flink/blob/master/flink-kubernetes/src/main/java/org/apa[…]/kubernetes/highavailability/KubernetesCheckpointIDCounter.java
LOG.info("Shutting down.");
if (jobStatus.isGloballyTerminalState()) {
LOG.info("Removing counter from ConfigMap {}", configMapName);
on termination, it removes the counter from the configmap that seems to be important to recover checkpoints.
Is this correct behaviour? (Flink 1.17.2 in our case)