Think of it as a system of low level K8s specific checks that is designed to be autonomous and does not need a third-party to run. Regardless of any node failu

Think of it as a system of low level K8s specific checks that is designed to be autonomous and does not need a third-party to run. Regardless of any node failures we can still get the diangostic information about cluster state (as long as there at least one node left). We actually plug in these checks to a higher-level monitoring and metrics system (in our case it's InfluxDB and not Nagios as you've suggested)