On a GKE cluster with cluster autoscaling, ensure every node runs a monitoring pod that sends container metrics to a third-party monitoring system. How should you deploy the pod?

A DaemonSet ensures that all (or some) nodes run a copy of a pod. This is ideal for cluster-level operations like monitoring agents, log collectors, or storage daemons, as it guarantees the monitoring pod will be present on every node, even as the cluster autoscales and adds new nodes. StatefulSets are used for stateful applications that require stable unique network identifiers, stable persistent storage, and ordered, graceful deployment and scaling. This is not applicable for a monitoring agent that needs to run on every node. Deployments are designed for stateless applications and manage a replicated set of pods. While you could create a Deployment, it wouldn't guarantee one pod per node and would require manual scaling or complex affinity rules to achieve the desired outcome, which is less efficient than a DaemonSet. Cluster initializers are a deprecated Kubernetes feature and are not used for deploying application pods. They were primarily for performing initial setup tasks on new clusters.