RPVST+ Problems between a 6509 and Marconi ESR5000

[Mike_Adkins]

I am trying to setup a network using Cisco 6509's and Marconi ESR5000's. I am running into a problem when I try and run spanning tree. The two different brands of switches are running the same version of STP. What’s happening is the root bridge is not being recognized between the two vendors, except for VLAN1. If I connect a 6509 and ESR5000 back to back they will recognize whoever the root bridge is for VLAN1 but all the other VLANs they see themselves as the root bridge. Even if I change the bridge priority. Any ideas?? I am wondering if it's a differnt BPDU multicast address between the different vendors. I am having trouble finding specific info on the details of each vendor.

[dnewell24]

Marconi does have a variant of PVST called dual layer which uses a different multicast address. Can you post the output of the 'show spantree version' on the ESR-5K and a 'show spantree' on the 6509?

[Francois Tallet]

I have absolutely no clue about the way the Marconi switch is operating, however, what you describe seems perfectly logical.

PVST is not an IEEE standard, it's a Cisco proprietary way of running one spanning tree per vlan. The IEEE only defines *one* spanning tree for all the vlans. Cisco PVST was created before 802.1Q that defined 1Q vlans. In order to be compatible with 802.1Q, it was slightly modified (and is now called PVST+). In PVST+, vlan 1 is *the* spanning tree interacting with the IEEE bridges. Vlan 1's BPDUs are sent untagged to the IEEE address. Other Vlans' BPDUs are sent to a cisco proprietary address that is flooded by third party switches. So basically, it is normal that your Marconi switch only receives Vlan 1's information. The Cisco PVST+ switches just see the third party network as a wire, but they still run STP over it. This should still prevent loops. If the Marconis are doing the same however, the result might not be very deterministic I'm afraid. The problem is that PVST+ was designed to interact with a standard IEEE bridge, not with another vendor doing the same hack as Cisco's;-)

If the Marconis are running some sort of PVST, the only way to get them to interact with the Cisco switches is to connect them with access ports. On access ports, BPDUs are always sent to the IEEE address by PVST+ switches (note that here, I make the assumption that the Marconis will do the same).

Finally, maybe the cleanest solution would be to run MST (if Marconi supports it). MST is the first IEEE protocol allowing several STP instance and should work with no problem (use the latest version on Cisco switches).

Regards,

Francois