Google's internal automation/service account has changed a firewall rule on our GKE cluster that blocked the pods from accessing hosts on the RFC1918 VPC the cluster was on (at 20:13:30 UTC). Did anyone experience something similar?