Metrics Server

Metrics Server is a lightweight Kubernetes component that collects container CPU and memory metrics from kubelets and exposes them through the Metrics API to…

  • Infrastructure
  • Tools & Utilities
Metrics Server is a lightweight Kubernetes component that collects container CPU and memory metrics from kubelets and exposes them through the Metrics API to power built-in Horizontal Pod Autoscaler and Vertical Pod Autoscaler pipelines, and to display resource usage via kubectl top for quick debugging.

Platform availability

Available on 1 of 10 platforms

  • Unraid (not listed)
  • TrueNAS (not listed)
  • Umbrel (not listed)
  • ZimaOS (not listed)
  • Proxmox (not listed)
  • Helm
  • Mac (not listed)
  • Windows (not listed)
  • Linux (not listed)
  • Any server (not listed)

Health score?A 0-100 score computed nightly from four factors: maintained (40%: recent commits and steady activity), popular (25%: GitHub stars, log scale), easy to install (20%: platforms, install notes, setup guides, Docker image) and light to run (15%: minimum RAM, ARM support). When the source code has been checked, a fifth factor, security check (20%: automated scan for committed secrets, unsafe desktop-app settings, known-vulnerable dependencies and project hygiene; any serious finding caps it at 20), is added. Factors without data are left out and the rest are rescaled.

77/100 Good

Maintained
100last commit 9 days ago
Popular
776,789 GitHub stars
Easy to install
251 install platform, install notes
Light to run
No data
Security check
84no serious findings, 2 minor, CI, tests, OpenSSF Scorecard 7.7/10

Based on 4 of 5 factors.

Availability: platform listings confirmed today

GitHub stars
6,789
Open issues
74
Last commit
2026-11-12
Latest release
v0.9.1
License
Apache-2.0
Activity
declining

Checked today - source: GitHub (kubernetes-sigs/metrics-server)

Resources & Compatibility

Requirements not published yet.

First-install notes

  • Pick a safe metrics port before starting If you use the Prometheus exporter, set the HTTP listener to an address you expect, because the simple example defaults to 0.0.0.0:9000 and lets any network host scrape metrics. If your metrics collector is only on the same machine, use 127.0.0.1:9000 or 9100 so outsiders cannot read them. source
  • Open only the exporter port you actually need If you use the TCP exporter, choose a listen_address such as 127.0.0.1:5000 and allow that TCP port only for the observer or monitoring tool that connects to it. Without the matching host and port, the observer has nothing to read and your metrics will not appear. source
  • Allow only trusted scrapers when possible For Prometheus scraping, add_allowed_address can limit which client IPs may request the metrics endpoint. This matters because an exposed metrics port can reveal app details or be probed by unwanted scanners on your network. source
  • Use only for autoscaling, not as a monitoring source The upstream README cautions that Metrics Server is meant only for autoscaling purposes: do not use it to forward metrics to monitoring solutions; collect metrics directly from the Kubelet /metrics/resource endpoint instead. source

Fetched 44 days ago - each note links to its source

Alternatives

No alternatives collected yet.

Common questions

How can I restrict who can query resource metrics?

Clients of the Kubernetes API can query the Metrics API. You can use Kubernetes access control mechanisms to manage permissions to do so.

Which Kubernetes autoscalers use Metrics Server data?

The HorizontalPodAutoscaler and VerticalPodAutoscaler use data from the Metrics API. They use it to adjust workload replicas and resources to meet demand.

Where does metrics-server collect its metrics from?

Metrics Server collects resource metrics from Kubelets. It exposes them through the Kubernetes apiserver using the Metrics API.

Is the Metrics API enough for full monitoring?

The Metrics API only offers the minimum CPU and memory metrics needed for automatic scaling with HPA and/or VPA. For a more complete set of metrics, you can complement it with a second pipeline using the Custom Metrics API.

Answers sourced from github.com, kubernetes.io

Community

Collected 48 days ago - public community threads