Configure Kubernetes Monitoring with Helm chart

Kubernetes Monitoring deploys the most recent version of the Grafana Kubernetes Monitoring Helm chart to your Cluster. To learn how the Helm chart configuration works, refer to Helm chart overview.

First, activate Kubernetes Monitoring in Grafana Cloud. Then choose a configuration method, run the Helm command on your Cluster, and confirm that data is flowing. Activation alone doesn’t start billing. Billing begins when your Cluster sends kube-state-metrics data.

Deploy the Helm chart using one of these methods:

Method	What it does	Choose this when
Configuration wizard	Walks you through deployment choices in Grafana Cloud, then generates the Helm values and deployment command for you to copy and run.	You want a guided setup focused on monitoring your Kubernetes infrastructure.
Manual values file	You write your own `values.yaml` file, and run the Helm install command.	You want full control over your values file from the start, or you manage deployments through your own scripts or a GitOps pipeline.

Available platforms

You can install Kubernetes Monitoring on these platforms:

Kubernetes, which includes Amazon Elastic Kubernetes Service (EKS) on Amazon Elastic Cloud Compute (EC2), Google Kubernetes Engine (GKE), and Tanzu
Amazon EKS on AWS Fargate
Azure Kubernetes Service (AKS)
Google Kubernetes Engine (GKE) Autopilot
IBM Cloud Kubernetes Service
Red Hat OpenShift

On every platform, the Helm chart scrapes kubelet and cAdvisor for container CPU and memory usage. Where Node CPU and memory usage comes from depends on the platform.

Platform	Source for Node CPU and memory usage
Kubernetes	Node Exporter, which the chart deploys
Amazon EKS on AWS Fargate	`kubelet` resource metrics. The chart doesn’t deploy Node Exporter because Fargate gives no access to the Node.
AKS	Node Exporter, which the chart deploys
GKE Autopilot	`kubelet` resource metrics. The chart doesn’t deploy Node Exporter because Autopilot doesn’t allow the privileged mounts it needs.
IBM Cloud	Node Exporter, which the chart deploys
OpenShift	The Node Exporter that OpenShift already runs in the `openshift-monitoring` namespace. The chart scrapes it, and the kube-state-metrics that OpenShift runs, instead of deploying its own.

Pod logs come from the log files on each Node, except on Amazon EKS on AWS Fargate, where the chart reads them through the Kubernetes API.

Activate and send data from your account

To configure and use Kubernetes Monitoring, you must activate the product and start sending data to your account. Refer to Activate and send data from your account.