These metrics can still be valuable to troubleshoot components running in cgroups on the host, outside the scope of Kubernetes pods.