CertGrid CertGrid
Troubleshooting·Certified Kubernetes Administrator

Resetting and Rejoining a Node with kubeadm

Take a node out of a cluster and put it back, and take a machine all the way back to bare Ubuntu. Covers what kubeadm reset does not clean up, why removing a node needs two commands rather than one, and the package purge that silently does nothing.

Installation and Cluster Builds Guide 13 of 103 Intermediate

Written against the versions above. Container and sandbox IDs, timings and package versions are specific to this run.

The worker was removed from this cluster and rejoined to it. The control plane stayed up throughout.
Server NameIP AddressOSRolesCPURAMHDD
CKA3001192.168.0.185Ubuntu 26.04 LTSControl Plane Node2 Core4 GB50 GB
CKA3001-NODE01192.168.0.186Ubuntu 26.04 LTSWorker Node (removed and rejoined)2 Core4 GB50 GB

Before you start

  1. Removing a node takes two commands, not one

  2. Reset the machine, and read what it tells you

  3. Expect noisy errors during reset on a node that had a CNI

  4. Rejoin with a fresh token

  5. Going further: back to bare Ubuntu

  6. The package purge that silently does nothing

  7. Reboot, because interfaces outlive the packages

Official sources