Our standard access closet configuration consists of a Catalyst switch in the access closet (layer 2 only), connected via two trunks -- one to each of two distribution switches which each have a layer 3 blade. In this particular case, all three switches are 6509's, and the layer 3 blades are MSFC2's. The MSFC's are configured with VLAN interfaces for the closet VLAN's and are set up for HSRP. The closet switch's bridge priorities have been modified so it becomes root for each of the VLAN's it services -- and both trunks would normally be forwarding. We recently installed a new 6509 with gig-e trunks to the dist switches. One of the trunks was showing physical errors and was causing slowness problems. We decided to disable the bad trunk port.
Given that the dist MSFC's use the access switch trunks (root bridge=access switch) for HSRP keepalives, what would be the expected HSRP behavior after bringing the bad trunk down? Our first thoughts were that there should be an HSRP swap, but that's not what happened. Instead, we lost routing to the HSRP'd VLAN's until the interfaces on the dist's of the bad trunk were manually brought down.
What happened? I suspect that because both HSRP'd VLAN interfaces were still up, even though HSRP keeaplives could not get through, both interfaces went ACTIVE. Are we expecting HSRP to work in this scenario when it really wasn't meant to?
HSRP provides high network availability because it routes IP traffic from hosts on Ethernet, FDDI, or Token Ring networks without relying on the availability of any single router. HSRP is used in a group of routers for selecting an active router and a standby router. (An active router is the router of choice for routing packets; a standby router is a router that takes over the routing duties when an active router fails, or when preset conditions are met.)
HSRP is supported over Inter-Switch Link (ISL) encapsulation,For more information :
Yes, both routers were HSRP active, but more importantly they both were advertising the same subnet that only one could really reach. So outbound packets from the access switch were OK, but the return packets were dying in the router with the bad link. Even when HSRP is fully functional return packets can go to either router. HSRP only serves to make 1 router the default gateway. And you can use GLBP to share the default gateway if you want to load balance in both directions.
A solution would be a trunk between the 2 distribution switches with the access vlans allowed. This would normally be blocking, but would provide fault tolerance for a link failure to the access switch.