The Kubernetes Resource Request That Quietly Wasted Half the Cluster

At 10:18, several production Pods stopped scheduling. Kubernetes returned: 0/64 nodes are available: 42 Insufficient cpu, 22 Insufficient memory. The message seemed inconsistent with what we were see