HSRP problems with trunking over a Cat5K

[bturbow]

Hello;

We are attempting to run HSRP as follows:

- We have 2 7507's trunked to 2 Cat5500's with GigE; where each 7507 is trunked to its respective Cat5K and the 2 Cat5K's are trunked together.

- A subinterface is created on each 7507 with dot1q encaps and has an address on a /24

- An HSRP group is configured on each 7507 within the /24

- Each 7507 can ping the other's IP on the /24 with no problem, and passing traffic between the routers works just fine.

- HSRP does not work. Performing a debug standby shows constant outgoing hello messages with nothing coming in. Here is the output from show standby:

GigabitEthernet5/0/0.100 - Group 100

Local state is Active, priority 110, may preempt

Hellotime 5 sec, holdtime 15 sec

Next hello sent in 3.228

Virtual IP address is 10.9.9.3 configured

Active router is local

Standby router is unknown

Virtual mac address is 0000.0c07.ac64

Authentication text "ts100"

4 state changes, last state change 00:22:32

IP redundancy name is "hsrp-Gi5/0/0.100-100" (default)

HSRP on each router simply does not see the other side's peer address.

I think the problem may be a multicast issue, where each side of the shared VLAN on each 7507 cannot see multicast from either side.

Thank you for any responses.

[bturbow]

OK, I was right. Everything is now working now that I've enabled multicast-routing and added "pim dense-mode" to each sub-interface.

New question: Do I have to add "pin dense-mode" to each sub-interface? It started working after I applied it on the 1st one. Is there another recommended way to enable multicast on the interfaces?

Comment: All over CCO and the docs it mentions multicast to support HSRP, but in NONE of the sample configs, etc. does it show multicast commands.

Thanks again for any response.

[Erick Bergquist]

PIM shouldn't be needed; Multicast packets should flow fine within the same IP Subnet (VLAN) assuming there is no problems at L2. PIM is needed when doing multicast between IP subnets (VLANs).

You probably were running into a HSRP Virtual MAC address versus switch CAM table issue. Since the vMAC is same for both routers switch might have issues keeping track of where this MAC is. Configuring HSRP to use bia (burned in address) fixes this problem.

Perhaps enabling PIM cleared something up somewhere. Does HSRP break when disabling PIM or continue to work?

Erick

[bturbow]

Yes, it breaks when PIM is removed.

Also, I tried standby use-bia and I get lots of "Defending ARP entry for 10.9.9.x" and duplicate address entries.

Also, I failed to mention in my original post that we are using GEChannel; there is no BIA on port-channel interfaces -- it shows all 0's in show int..

So, I'm OK with using PIM but I think there may be a cleaner way to allow multicast to go accross the Gig channels through the Cats to the other 7507.

Thanks again.