One of our upstream providers is performing emergency network maintenance on a subset of hosts in ORD. Some machines in ORD may be unreachable during the incident. A small subset of MPG clusters are running with reduced HA due to their replicas being on the affected hosts.