I'm currently using a prometheus statefulset down rule to monitor our statefulsets and have noticed the alert is firing incorrectly for the elasticsearch statefulsets managed by ECK. It seems to be caused by updates. The statefulset currentRevision is not being updated because the OnDelete updateStrategy is used, which causes kube-state-metrics to return the wrong result.
Should the ECK operator update the current revision of its statefulsets once it has restarted all pods within it?