Answer
Check the pod's own event/description output for a resource-scheduling reason before investigating the application layer — a pending pod very often means the cluster doesn't currently have enough CPU or memory available to schedule it, which is a cluster capacity question, not an application defect.