Check manual page of kube_replicas
|Distribution||official part of Check_MK|
In the event that the number of ready or up-to-date replicas is below the number of desired replicas, the check shows a WARN and CRIT state if this persists for more than 5 minutes and 10 minutes respectively. Moreover, if there are misscheduled Pods controlled by a DaemonSet, then the check will also show WARN/CRIT, after 5/10 minutes.
It is possible to configure the grace period for each state using the "Kubernetes: Replicas" ruleset.
Please note that replicas may temporarily be in a not ready, out-of-date or misscheduled state during the process of an update. It is therefore advised to always have a grace period configured.