How to fix unable to compute replica count: the metrics for metric was not available in Kubernetes

KubernetesINTERMEDIATEHIGH

HorizontalPodAutoscaler fails to scale because the metric it references does not exist or is invalid. Metrics may be missing from metrics-server or custom metrics provider.

What this error means

HPA uses metrics to decide how many pod replicas to create. When the metric is invalid, does not exist, or has no data, HPA cannot compute the desired replica count. This can be because the metrics-server is not installed, the metric name is incorrect, or the metric is not being exported by your application.

Typical symptoms

HPA status shows "unknown" for current/desired replicas
HPA events show "unable to compute replica count"
Metric names do not match what metrics-server provides
Custom metrics not available to HPA
kubectl describe hpa shows "Metrics: (unknown)"
HPA scaling does not trigger despite high resource usage
Pod metrics visible in metrics API but HPA cannot access

Common causes

metrics-server not installed or running in cluster
Metric name incorrect (e.g., "cpu" vs "cpu_usage")
Custom metrics not registered with metrics API
Metric missing labels or incorrect label selectors
Metrics data not available yet (fresh deployment)
Pod missing resource requests (required for CPU metric)
Metrics-server RBAC permissions insufficient

How to fix "unable to compute replica count: the metrics for metric was not available"

1Step 1

First diagnostic step

2Step 2

Second diagnostic step

3Step 3

Third diagnostic step

4Step 4

Fourth diagnostic step

Step 5

Advanced notes

Additional notes and platform-specific considerations.

Related errors

Official resources & further reading

📚Kubernetes Official Documentation
📖
Kubernetes Docs: Horizontal Pod Autoscaling(Official Docs)
📖
Kubernetes Docs: HPA Walkthrough(Official Docs)
💬
Stack Overflow: Kubernetes HPA metrics(Stack Overflow)