kubeadm reset

Performs a best effort revert of changes made by kubeadm init or kubeadm join.

Synopsis

Performs a best effort revert of changes made to this host by 'kubeadm init' or 'kubeadm join'

The "reset" command executes the following phases:

preflight           Run reset pre-flight checks
remove-etcd-member  Remove a local etcd member.
cleanup-node        Run cleanup node.

kubeadm reset [flags]

Options

Options inherited from parent commands

Reset workflow

kubeadm reset is responsible for cleaning up a node local file system from files that were created using the kubeadm init or kubeadm join commands. For control-plane nodes reset also removes the local stacked etcd member of this node from the etcd cluster.

kubeadm reset phase can be used to execute the separate phases of the above workflow. To skip a list of phases you can use the --skip-phases flag, which works in a similar way to the kubeadm join and kubeadm init phase runners.

kubeadm reset also supports the --config flag for passing a ResetConfiguration structure.

Cleanup of external etcd members

kubeadm reset will not delete any etcd data if external etcd is used. This means that if you run kubeadm init again using the same etcd endpoints, you will see state from previous clusters.

To wipe etcd data it is recommended you use a client like etcdctl, such as:

etcdctl del "" --prefix

See the etcd documentation for more information.

Cleanup of CNI configuration

CNI plugins use the directory /etc/cni/net.d to store their configuration. The kubeadm reset command does not cleanup that directory. Leaving the configuration of a CNI plugin on a host can be problematic if the same host is later used as a new Kubernetes node and a different CNI plugin happens to be deployed in that cluster. It can result in a configuration conflict between CNI plugins.

To cleanup the directory, backup its contents if needed and then execute the following command:

sudo rm -rf /etc/cni/net.d

Cleanup of network traffic rules

The kubeadm reset command does not clean any iptables, nftables or IPVS rules applied to the host by kube-proxy. A control loop in kube-proxy ensures that the rules on each node host are synchronized. For additional details please see Virtual IPs and Service Proxies.

Leaving the rules without cleanup should not cause any issues if the host is later reused as a Kubernetes node or if it will serve a different purpose.

If you wish to perform this cleanup, you can use the same kube-proxy container which was used in your cluster and the --cleanup flag of the kube-proxy binary:

docker run --privileged --rm registry.k8s.io/kube-proxy:v1.34.0 sh -c "kube-proxy --cleanup && echo DONE"

The output of the above command should print DONE at the end. Instead of Docker, you can use your preferred container runtime to start the container.

Cleanup of $HOME/.kube

The $HOME/.kube directory typically contains configuration files and kubectl cache. While not cleaning the contents of $HOME/.kube/cache is not an issue, there is one important file in the directory. That is $HOME/.kube/config and it is used by kubectl to authenticate to the Kubernetes API server. After kubeadm init finishes, the user is instructed to copy the /etc/kubernetes/admin.conf file to the $HOME/.kube/config location and grant the current user access to it.

The kubeadm reset command does not clean any of the contents of the $HOME/.kube directory. Leaving the $HOME/.kube/config file without deleting it, can be problematic depending on who will have access to this host after kubeadm reset was called. If the same cluster continues to exist, it is highly recommended to delete the file, as the admin credentials stored in it will continue to be valid.

To cleanup the directory, examine its contents, perform backup if needed and execute the following command:

rm -rf $HOME/.kube

Graceful kube-apiserver shutdown

If you have your kube-apiserver configured with the --shutdown-delay-duration flag, you can run the following commands to attempt a graceful shutdown for the running API server Pod, before you run kubeadm reset:

yq eval -i '.spec.containers[0].command = []' /etc/kubernetes/manifests/kube-apiserver.yaml
timeout 60 sh -c 'while pgrep kube-apiserver >/dev/null; do sleep 1; done' || true

What's next

• kubeadm init to bootstrap a Kubernetes control-plane node
• kubeadm join to bootstrap a Kubernetes worker node and join it to the cluster