Resolving Kubelet Startup Failure Caused by Mismatched NTO Image During Upgrade
Issue
During an OpenShift upgrade from 4.16.42 to 4.17.32, a master node failed to start the kubelet and crio services after a reboot. As a result, the OpenShift upgrade became stuck and did not progress.
The node's ocp-tuned-one-shot.service repeatedly failed due to a mismatch between the configured cluster-node-tuning-operator image and the version expected by the upgrade process.
You may see the below error in the journal log.
Error: unknown command "openshift-tuned" for "cluster-node-tuning-operator"
Environment
- OpenShift Container Platform 4.16
Subscriber exclusive content
A Red Hat subscription provides unlimited access to our knowledgebase, tools, and much more.