You created a Deployment with 2 replicas on a GKE cluster with a single preemptible node pool; one Pod remains Pending. What is the most likely cause?

The most likely cause for a Pod remaining in a Pending state in a GKE cluster with a single preemptible node pool is insufficient available resources. Preemptible VMs are often used for cost savings and can be reclaimed by Google Cloud, but even without preemption, if existing Pods consume all available CPU or memory on the nodes, new Pods cannot be scheduled until resources free up or new nodes are added (which won't happen automatically with a fixed-size node pool). The Pod's resource requests exceeding a single node's capacity is less likely to be the most likely cause, as Kubernetes typically tries to schedule Pods on nodes that can accommodate them. A service account lacking image pull permissions would result in an ImagePullBackOff status, not Pending. While preempted nodes can cause rescheduling, the question states one Pod remains Pending, implying a persistent scheduling issue rather than a temporary preemption event.