Explain how you would upgrade a production Kubernetes cluster with zero downtime for Baker Hughes workloads.
