The teardown removes everything the bootstrap installed. Use it to start over in a lab, or to rehearse a total loss.
#Run it
From the controller:
make teardownAnsible asks you to type destroy before it touches anything. Any other answer cancels the run.
#What it removes
On every node:
- RKE2, through its own uninstall script
/var/lib/rancher/rke2and/etc/rancher/rke2, which hold etcd and the cluster configuration- the
kubectl,crictlandctrlinks and the login profile - the CIS kernel settings file
/var/lib/longhorn, every volume replica on the node
On the controller, it deletes ~/.bedrock. That holds more than the admin kubeconfig: OpenBao’s unseal shares and root token, the failover proof keys and the observers’ exchange certificate authority. It only runs once every node has been wiped. If any node fails, the teardown stops there and the controller keeps the keys, because those nodes still hold data the keys belong to.
#Deploy again
The nodes are ready for a fresh bootstrap straight away:
make deploymake redeploy runs the teardown and the deploy in one go.