Skip to main content

Pod Metrics

KubeVision displays current CPU and memory usage in Workloads → Pods. The list shows totals for each Pod and refreshes every 15 seconds while it is open.

Open a Pod and use the Overview tab to see:

  • Total CPU and memory usage
  • Configured requests and limits
  • Usage as a percentage of the configured limit
  • CPU and memory usage for each regular container

Requirements​

The cluster must expose the Kubernetes Metrics API through Metrics Server at metrics.k8s.io/v1beta1. The credentials used to connect the cluster need get and list access to pods in that API group. The KubeVision Helm chart grants only those read operations to its in-cluster service account.

If Metrics Server is starting, a new Pod may briefly show that metrics are pending. If the API is unavailable or access is denied, the Pods page remains usable and shows an unavailable state instead of failing the resource request.

Metrics Server does not report container filesystem or PVC usage. Those values require a monitoring source such as kubelet/cAdvisor or Prometheus and are not estimated from the node filesystem shown by df inside a container.