You want your Google Kubernetes Engine cluster to automatically add or remove nodes based on CPU load. What should you do?

To automatically scale nodes based on CPU load, you need two components working together. First, the HorizontalPodAutoscaler (HPA) monitors CPU utilization of your pods and scales the number of pods up or down within the cluster. Second, the Cluster Autoscaler (CA) monitors resource requests and actual usage across the cluster. If there aren't enough nodes to schedule new pods (scaled by HPA) or if nodes are underutilized, the CA adds or removes nodes from the node pools. Enabling the Cluster Autoscaler from the GCP Console is the standard way to activate this feature for GKE. The other options either misconfigure the autoscaling mechanisms or use deployment properties that are unrelated to node autoscaling.