‹ All CKA questions

CKA Practice Question 2: kubeadm Worker Node Upgrade

Certified Kubernetes Administrator (CKA) · Test 1 · question 2 of 10

Cluster Upgradesmedium

You are upgrading a kubeadm-built cluster one minor version. The control plane nodes have already been upgraded successfully and are serving traffic. You now need to upgrade the worker nodes with the least disruption to running workloads.

Which sequence should you follow on each worker node?

Select one answer
Show answer and explanation

Answer:

  • A. Upgrade the kubeadm package, run kubeadm upgrade node, drain the node, upgrade kubelet and kubectl, restart the kubelet, then uncordon the node

This is the sequence documented by kubeadm for worker nodes. kubeadm upgrade node updates the node's local kubelet configuration, draining evicts workloads before the kubelet restarts, and uncordoning returns the node to the schedulable pool afterwards.

Option B uses the wrong subcommand. kubeadm upgrade apply is run once, on the first control plane node, to upgrade the cluster's control plane components; on a worker it is not the correct operation. It also skips draining entirely, so pods are killed abruptly when the kubelet restarts.

Option C would eventually produce an upgraded node, but deleting and rejoining a node is far more disruptive than an in-place upgrade and discards the node's identity and any node-local state; it is not the supported upgrade path.

Option D inverts the order. kubeadm upgrade node writes the kubelet configuration that the new kubelet is expected to read, so upgrading and restarting the kubelet beforehand leaves it running briefly against stale configuration.

Reference: Upgrading kubeadm clusters

Practice Test 1 ›

Study guides for CKA