What Can I Do If a Node Is Not Ready After Its Scale-Out?

Symptom

After a node in the on-premises cluster is scaled out, the Kubernetes resources of the node may not be able to be started, that is, the node is not ready.

Procedure

The /mnt/paas/kubernetes/kubelet/cpu_manager_state file stores the original cpu_manager_policy, which is the core binding setting of the original CPU cores and needs to be deleted. Restart kubelet to enable cpu_manager to bind cores based on the existing CPU topology and generate cpu_manager_state again.

Run the following commands:

rm /mnt/paas/kubernetes/kubelet/cpu_manager_state

systemctl restart kubelet

Wait for a period of time until the node is working.