Kubelet is not a control plane component, but it plays a key role in the operation of a Kubernetes cluster.

Kubelet is an agent that runs on every node in a Kubernetes cluster. It ensures that containers in pods are started and run according to their specifications. Kubelet continuously interacts with kube-apiserver to verify and maintain the state of nodes and containers. It is also responsible for starting control plane components.

Static pod manifests

Kubelet starts control plane components from static pod manifests located in the /etc/kubernetes/manifests directory. In Deckhouse Kubernetes Platform, kubelet processes only files with the .yaml or .yml extension in this directory.

Files with other extensions, such as kube-apiserver.backup, kube-apiserver.yaml.bak, editor swap files, or other temporary files, are ignored. This prevents accidental processing of backup or non-manifest files as static pod manifests.

Kubelet interactions

Kubelet interactions are shown in the architecture diagram of the control-plane-manager module.

Kubelet interacts with the following components:

  1. kubernetes-api-proxy: Proxies requests to kube-apiserver that are sent to the localhost address. It is part of the control-plane-manager module.
  2. kube-apiserver-healthcheck: Checks the health of kube-apiserver.

The following components interact with kubelet:

  1. kube-apiserver:

    • Retrieving pod logs (the kubectl logs command)
    • Executing commands in running pods (the kubectl exec command)
    • Port forwarding (the kubectl port-forward command)

  2. prometheus-main: Collects kubelet metrics.

Additional resources