Measured CPU load × 12 W/core, with 85 % of server heat captured for hot water — the Leafcloud model — over the last 24 hours. History lives in a ConfigMap, so it survives pod restarts and hibernation. An estimate, honestly labelled.
Burns CPU in the app pods for 60 s. The HPA scales replicas at 60 % CPU; push far enough and the cluster autoscaler asks Nova for a new node. Watch the node cards and events above.