Pod affinity rules are too strict, preventing pods from scheduling on available nodes even when resources are free. This causes resource starvation and degraded service. Relaxing affinity labels and testing alternate node selectors will restore normal scheduling.