I recently inherited a production network that splits a single /24 across two vlans. I have 2 Juniper M20’s with 2 foundry fastirons hanging off of them. I load balance to the web service via 2 server irons. The previous admin, and architect, created an admin/routed vlan and a load balanced vlan. I suffer from strange reacahbility issues. i.e. unable to tftp from load balancers to tftp service on an admin vlan box. Until I refresh arp caches via ping. I have recieved push-back on redesigning this…. Honestly I’m not even sure how this works, shouldn’t the hosts on separate subnets be completely unable to reach eachother? There isn’t any port overlap or layer3 vlans configured. Is this a “one-armed routing solution”? If so shouldn’t I see the border routers appear as a hop on any traceroute between any boxes on separate vlans???