Cluster API (CAPI)

Overview

Cluster API (CAPI) manages Kubernetes clusters as Kubernetes resources. Instead of SSH-ing into nodes and running imperative commands, you define clusters declaratively using Custom Resources, and CAPI controllers reconcile the desired state into reality.

Core principle: CAPI turns cluster lifecycle management into a Kubernetes-native workflow -- define a Cluster CR, apply it, and controllers handle provisioning, scaling, and upgrades automatically.

When to Use

• Provisioning Kubernetes clusters declaratively from a management cluster
• Managing fleet of clusters at scale (tens to hundreds)
• Performing rolling upgrades by patching a version field
• Using ClusterClass blueprints for standardized cluster templates
• Setting up dev/test environments with the Docker infrastructure provider

Not for: Single cluster installation with security hardening (use rke2-deployment or kubespray-deployment), air-gapped environments without CAPI controller images, bare-metal provisioning without an infrastructure provider

Key Concepts