Skip to main content

Cluster API core concepts

Intro​

Via the official docs:

Cluster API is a Kubernetes sub-project focused on providing declarative APIs and tooling to simplify provisioning, upgrading, and operating multiple Kubernetes clusters.

Started by the Kubernetes Special Interest Group (SIG) Cluster Lifecycle, the Cluster API project uses Kubernetes-style APIs and patterns to automate cluster lifecycle management for platform operators. The supporting infrastructure, like virtual machines, networks, load balancers, and VPCs, as well as the Kubernetes cluster configuration are all defined in the same way that application developers operate deploying and managing their workloads. This enables consistent and repeatable cluster deployments across a wide variety of infrastructure environments.

Cluster types​

Management Cluster​

A Management cluster manages the state and lifecycle of Workload clusters using components called providers. Each Management cluster stores and reconciles the Cluster API resources (e.g. Machine, MachineDeployment, etc.) of Workload clusters by running one or more providers.

Workload Cluster​

Workload clusters, as the name suggests, are used to run and orchestrate the application workloads of the user. Workload clusters, in the context of Cluster API, are always managed by a Management cluster.

Providers​

Infrastructure Provider​

Infrastructure providers are responsible for provisioning the necessary infrastructure and compute resources. Each node, regardless of its type (e.g. a VM or baremetal), requires specific configuration options which these providers use during the provisioning process e.g. OS image and checksum, network settings, etc.

A popular and widely adopted baremetal infrastructure provider is the CAPM3 project (Cluster API Provider Metal³). It enables users to deploy a Cluster API based cluster using Metal3.

Bootstrap Provider​

Bootstrap providers are responsible for turning a fully provisioned server into a Kubernetes node. This includes, but is not limited to, configuring, initializing and joining control plane and worker nodes, generating kubeconfig and cluster certificates, etc.

The CAPRKE2 project (Cluster API Provider RKE2) aims to provide both Control Plane and Bootstrap providers for RKE2 based clusters. It is currently in early development by the Rancher team.