derailed/k9s · warning

no metrics-server detected on cluster

Error message

no metrics-server detected on cluster

What it means

Returned by MetricsServer.checkAccess (internal/client/metrics.go:92) when HasMetrics() reports the cluster does not expose the metrics.k8s.io API. HasMetrics() delegates to supportsMetricsResources(), a discovery call checking for the metrics-server aggregated APIService. The error fires before any RBAC check, on every FetchNodesMetrics/FetchNodeMetrics/FetchPodsMetrics/FetchPodMetrics call, which all funnel through checkAccess.

Source

Thrown at internal/client/metrics.go:92

			nodeMetrics[nmx.Items[i].Name] = node
		}
	}

	var ccpu, cmem, tcpu, tmem int64
	for _, mx := range nodeMetrics {
		ccpu += mx.CurrentCPU
		cmem += mx.CurrentMEM
		tcpu += mx.AllocatableCPU
		tmem += mx.AllocatableMEM
	}
	mx.PercCPU, mx.PercMEM = ToPercentage(ccpu, tcpu), ToPercentage(cmem, tmem)

	return nil
}

func (m *MetricsServer) checkAccess(ns string, gvr *GVR, msg string) error {
	if !m.HasMetrics() {
		return errors.New("no metrics-server detected on cluster")
	}

	auth, err := m.CanI(ns, gvr, "", ListAccess)
	if err != nil {
		return err
	}
	if !auth {
		return errors.New(msg)
	}
	return nil
}

// NodesMetrics retrieves metrics for a given set of nodes.
func (*MetricsServer) NodesMetrics(nodes *v1.NodeList, metrics *mv1beta1.NodeMetricsList, mmx NodesMetrics) {
	if nodes == nil || metrics == nil {
		return
	}

View on GitHub (pinned to 2d3ccc6ba2)

Solutions

1. Install metrics-server in the cluster and wait for APIService v1beta1.metrics.k8s.io to become Available (kubectl get apiservices)
2. Check the metrics-server pod logs and readiness: kubectl -n kube-system get pods | grep metrics-server
3. If RBAC hides the group, grant list/get on apiservices or fix the aggregation-layer certs
4. Gracefully degrade: treat this error as 'metrics feature unavailable' instead of a hard failure

Example fix

// before
mx, err := metricsSrv.FetchNodesMetrics(ctx)

// after
if !metricsSrv.HasMetrics() {
    return nil, nil // metrics not deployed; skip pulse view gracefully
}
mx, err := metricsSrv.FetchNodesMetrics(ctx)

if !metricsSrv.HasMetrics() {
    return nil, nil // or a typed 'metrics unavailable' sentinel
}
mx, err := metricsSrv.FetchNodesMetrics(ctx)

mx, err := metricsSrv.FetchNodesMetrics(ctx)
if err != nil && strings.Contains(err.Error(), "no metrics-server detected") {
    // render UI without metrics columns; do not treat as an error dialog
}

When it happens

Trigger: Calling any metrics fetch (FetchNodesMetrics at metrics.go:155/197, FetchPodsMetrics at :237, FetchPodMetrics at :299) on a cluster without metrics-server installed, with the aggregated APIService unavailable, or while discovery is still populating after cluster startup.

Common situations: Minimal/local clusters (kubeadm without addons, kind without the metrics-server addon), metrics-server pod crash-looping or not yet ready, APIService v1beta1.metrics.k8s.io Available=False, RBAC blocking discovery of the API group so it appears absent.

Related errors

• user is not authorized to list node metrics
• user is not authorized to list pods metrics
• user is not authorized to list pod metrics
• ACCESS -- No API server connection
• no connection to cached dial

AI-assisted analysis of derailed/k9s@2d3ccc6ba2 (2026-08-15). Data as JSON: /api/errors/ca3987cac0281820. Report an issue: GitHub.