We provide internet to several locations in town. We currently have a 10G internet connection with a /23 that comes into a Cisco 6505. We carve that up into /28 and /29 subnets and vlan them out to our different properties in town. We have a 10Gb connection to the city fiber network and 1Gb connections from the city fiber network to each property. Bandwidth at each property is configured for 200Mb down, burstable to 1Gb. The city network is configured as a full mesh. At each property we have a Ubiquity Edge-Router Pro for the edge device. We have the Edge-Router doing NAT for 40 to 400 clients. There is a couple bridges configured on the Edge-Routers, 1 to bring a vlan from the 6506 for any clients that has a public IP address and another one for a management VLAN for our Cisco switches on site. Each Edge-Router is then connected to up to 20 Cisco 3550 or 3560 switches to feed each client. Each client will have a 100Mb connection.
The issue we are having is at random times the internet will drop out. We can run a continuous ping while connected to a client side and we are not able to ping the router. Pings on the management vlan have no issues. We have run wireshark and we dont see any issues. No massive amounts of broadcast to indicate a broadcast storm. The only weird thing is we see a bunch of ARP requests coming from the router during the timeouts. The Cisco switches are configured with port protected mode, STP portfast and bpduguard. No timeout issues with the 6506. Any one have an idea?