Skip to content

Metrics ​

When the cluster runs metrics-server, the Pod, Deployment, StatefulSet and DaemonSet tables get CPU and Memory columns in kubectl top style (a workload sums its pods), and Overview shows usage against the summed requests and limits. A pod or workload at 80 % or more of a CPU or memory limit gets a cpu 85% / mem 92% badge on the graph. Usage is sampled every 15 s and never changes an object's status. Without metrics-server, or without access to pod metrics, Overview says so.

To try it locally, run the demo with --with-metrics (see Try it on a demo cluster).