Question

A cluster administrator needs to perform a firmware update on a GPU worker node in a Kubernetes cluster running NVIDIA GPU workloads. Several inference pods are currently scheduled on that node. The administrator wants to safely remove running pods and prevent new pods from being scheduled there before starting maintenance, while allowing the workloads to be rescheduled elsewhere. Which action should the administrator take?

Accepted Answer

Run 'kubectl cordon' followed by 'kubectl drain' on the node

Answer

Delete the node object from the cluster with 'kubectl delete node'

Answer

Scale the GPU Operator deployment to zero replicas

Answer

Manually kill the containerd processes on the affected node