Skip to content

Metric Server

Overview

The Kubernetes Metrics Server is a cluster-wide aggregator of resource usage data, such as CPU and memory, used by the Kubernetes control plane for autoscaling and monitoring. In KubeDNA, the Metrics Server is installed by default with every Kubernetes cluster to enable core functionality like the Horizontal Pod Autoscaler and real-time resource visualization.

✅ Default Installation

In KubeDNA, the Metrics Server is automatically deployed as part of the base system components.

  • No manual installation is required.
  • Deployed in the kube-system namespace.
  • Integrated with the Kubernetes API to expose resource metrics via the metrics.k8s.io API.
  • Fully compatible with HPA, Keda, and other monitoring tools.

🔧 Functionality Provided

  • Pod-level resource metrics (CPU and memory)
  • Node-level metrics for cluster overview
  • Support for Horizontal Pod Autoscaler (HPA)
  • Real-time resource graphs within the KubeDNA dashboard
  • Enables dashboards and observability tools (Grafana, Prometheus, etc.) to retrieve live metrics

🔍 Use Cases

  • Autoscaling workloads based on live CPU/memory usage
  • Observing performance bottlenecks per pod or node
  • Triggering alerts or scaling policies using external tools (e.g., Keda, Prometheus)
  • Enhancing capacity planning and debugging

📌 Notes

  • Metrics are collected at short intervals and stored temporarily in memory (not long-term).
  • For historical metrics or custom metrics, you can integrate with Prometheus and OpenTelemetry, both supported in KubeDNA.
  • The Metrics Server is automatically monitored and restarted if health checks fail, as part of KubeDNA’s self-healing platform design.