Question

A worker node named worker-3 shows status NotReady. You SSH into the node and find the kubelet is running. Running 'kubectl describe node worker-3' shows the condition 'DiskPressure' is True and 'Ready' is False with the message 'kubelet has disk pressure'. Existing pods on the node are being evicted. Which action correctly diagnoses and resolves the root cause?

Accepted Answer

Free up disk space on the node's image and container filesystem (e.g., prune unused images/logs) so usage drops below the kubelet eviction threshold, allowing the DiskPressure condition to clear.

Answer

Restart the kubelet service with 'systemctl restart kubelet' to reset the node conditions and mark it Ready again.

Answer

Delete and recreate the CNI plugin pods on the node, since DiskPressure is caused by a networking backlog.

Answer

Increase the pod eviction timeout in the kube-controller-manager so pods are no longer evicted from worker-3.