- Static pods are managed directly by the Kubelet daemon.
- The API server does not observe static pods. This means that static
pods are independent of all control plane components (other than the
kubelet) and thus can be started (and are usually started) before the
cluster itself.
- The Kubelet does however create a mirror pod on the Kubernetes API
server for each static pod, allowing them to be visible on the API
server, but cannot be controlled from there. The pod names will be
suffixed with the node’s hostname(“control” in the example below) with
a leading hyphen.
NAMESPACE NAME READY STATUS
kube-system etcd-control 1/1 Running
kube-system kube-apiserver-control 1/1 Running
kube-system kube-controller-manager-control 1/1 Running
kube-system kube-scheduler-control 1/1 Running
kubeadm init spins up 4 static pods - etcd, apiserver, controller &
scheduler in the kube-system namespace. Since these components are
what make up the control plane, they need to run as static pods.- You most likely would never have to deal with static pods directly,
unless you’re deploying a high availability cluster using
kube-vip or
similar.