Kubernetes Metrics Server

Metrics Server is a scalable, efficient source of container resource metrics for Kubernetes built-in autoscaling pipelines.

Metrics Server collects resource metrics from Kubelets and exposes them in Kubernetes apiserver through Metrics API for use by Horizontal Pod Autoscaler and Vertical Pod Autoscaler. Metrics API can also be accessed by kubectl top, making it easier to debug autoscaling pipelines.

Metrics Server is not meant for non-autoscaling purposes. For example, don't use it to forward metrics to monitoring solutions, or as a source of monitoring solution metrics. In such cases please collect metrics from Kubelet /metrics/resource endpoint directly.

Metrics Server offers:

• A single deployment that works on most clusters (see Requirements)
• Fast autoscaling, collecting metrics every 15 seconds.
• Resource efficiency, using 1 mili core of CPU and 2 MB of memory for each node in a cluster.
• Scalable support up to 5,000 node clusters.

Use cases

You can use Metrics Server for:

• CPU/Memory based horizontal autoscaling (learn more about Horizontal Autoscaling)
• Automatically adjusting/suggesting resources needed by containers (learn more about Vertical Autoscaling)

Don't use Metrics Server when you need:

• Non-Kubernetes clusters
• An accurate source of resource usage metrics
• Horizontal autoscaling based on other resources than CPU/Memory

For unsupported use cases, check out full monitoring solutions like Prometheus.

Requirements

Metrics Server has specific requirements for cluster and network configuration. These requirements aren't the default for all cluster distributions. Please ensure that your cluster distribution supports these requirements before using Metrics Server:

• Metrics Server must be reachable from kube-apiserver by container IP address (or node IP if hostNetwork is enabled).
• The kube-apiserver must enable an aggregation layer.
• Nodes must have Webhook authentication and authorization enabled.
• Kubelet certificate needs to be signed by cluster Certificate Authority (or disable certificate validation by passing --kubelet-insecure-tls to Metrics Server)
• Container runtime must implement a container metrics RPCs (or have cAdvisor support)

Installation

Latest Metrics Server release can be installed by running:

kubectl apply -f https://github.com/kubernetes-sigs/metrics-server/releases/latest/download/components.yaml

Installation instructions for previous releases can be found in Metrics Server releases.

Compatibility matrix:

Metrics Server	Metrics API group/version	Supported Kubernetes version
0.6.x	metrics.k8s.io/v1beta1	*1.19+
0.5.x	metrics.k8s.io/v1beta1	*1.8+
0.4.x	metrics.k8s.io/v1beta1	*1.8+
0.3.x	metrics.k8s.io/v1beta1	1.8-1.21

*For <1.16 requires passing --authorization-always-allow-paths=/livez,/readyz command line flag

Security context

Metrics Server requires the CAP_NET_BIND_SERVICE capability in order to bind to a privileged ports as non-root. If you are running Metrics Server in an environment that uses PSPs or other mechanisms to restrict pod capabilities, ensure that Metrics Server is allowed to use this capability. This applies even if you use the --secure-port flag to change the port that Metrics Server binds to to a non-privileged port.

Scaling

Starting from v0.5.0 Metrics Server comes with default resource requests that should guarantee good performance for most cluster configurations up to 100 nodes:

• 100m core of CPU
• 200MiB of memory

Metrics Server resource usage depends on multiple independent dimensions, creating a Scalability Envelope. Default Metrics Server configuration should work in clusters that don't exceed any of the thresholds listed below:

Quantity	Namespace threshold	Cluster threshold
#Nodes	n/a	100
#Pods per node	70	70
#Deployments with HPAs	100	100

Resources can be adjusted proportionally based on number of nodes in the cluster. For clusters of more than 100 nodes, allocate additionally:

• 1m core per node
• 2MiB memory per node

You can use the same approach to lower resource requests, but there is a boundary where this may impact other scalability dimensions like maximum number of pods per node.

Configuration

Depending on your cluster setup, you may also need to change flags passed to the Metrics Server container. Most useful flags:

• --kubelet-preferred-address-types - The priority of node address types used when determining an address for connecting to a particular node (default [Hostname,InternalDNS,InternalIP,ExternalDNS,ExternalIP])
• --kubelet-insecure-tls - Do not verify the CA of serving certificates presented by Kubelets. For testing purposes only.
• --requestheader-client-ca-file - Specify a root certificate bundle for verifying client certificates on incoming requests.

You can get a full list of Metrics Server configuration flags by running:

docker run --rm k8s.gcr.io/metrics-server/metrics-server:v0.5.0 --help

Helm Chart

This Helm chart can deploy the metric-server service in your cluster.

Note: This Helm chart isn't supported by Metrics Server maintainers.

Design

Metrics Server is a component in the core metrics pipeline described in Kubernetes monitoring architecture.

For more information, see:

• Metrics API design
• Metrics Server design

Have a question?

Before posting an issue, first checkout Frequently Asked Questions and Known Issues.

Community, discussion, contribution, and support

Learn how to engage with the Kubernetes community on the community page.

You can reach the maintainers of this project at:

• Slack channel
• Mailing list

This project is maintained by SIG Instrumentation

Code of conduct

Participation in the Kubernetes community is governed by the Kubernetes Code of Conduct.