[c-nsp] Problems manipulating rapid spanning tree

Sam Stickland sam_ml at spacething.org
Thu Aug 5 08:03:14 EDT 2004


Hi,

I've got the following network layout

      +--  A: cisco 6509 --+
      |                    |
      FE                   |
      |                    |
  B:cisco 3550             GE
      |                    |
      FE                   |
      |                    |
  C:cisco 2950 ---GE--- D:cisco 2950


Switch A is the primary root of the spanning tree domain for vlan 218. 
Switches C and D are access switches and have uplinkfast enabled.

On switch C, 'sh spanning-tree vlan 218' shows the following:

Interface        Role Sts Cost      Prio.Nbr Type
---------------- ---- --- --------- -------- ----
Fa0/4            Desg FWD 3019      128.4    P2p
FE               Root FWD 3019      128.25   P2p
GE               Altn BLK 3004        0.26   P2p

(I've changed the interface names to fit in with the speed designators in 
the diagram).

Why is switch C choosing to take the FE (100Mbit) paths to the primary 
root, instead of the the GE paths?

As you can see I've also tried setting the port priority on switch C's GE 
port to 0 to try and influence a tie-breaking decision. The path cost is 
also listed as less (3004 versus 3019).

For the sake of completeness here's vlan 218's spanning-tree as switches 
A, B and D see it.

Switch A (cisco 6509)
---------------------

Interface        Role Sts Cost      Prio.Nbr Type
---------------- ---- --- --------- -------- ----
GE               Desg FWD 4         128.2    P2p
FE               Desg FWD 19        128.193  P2p



Switch B (cisco 3550)
---------------------

Interface   Port ID                   Designated                Port ID
Name        Prio.Nbr    Cost Sts      Cost Bridge ID            Prio.Nbr
----------- -------- ------- --- --------- -------------------- --------
FE(C)       128.18        19 FWD        19 32986 000b.4636.8e80 128.18
FA(A)       128.21        19 FWD         0 24794 00b0.c23b.7c00 128.193




Switch D (cisco 2950)
---------------------

Interface        Role Sts Cost      Prio.Nbr Type
---------------- ---- --- --------- -------- ----
GE(A)            Root FWD 3004      128.25   P2p
GE(C)            Desg FWD 3004      128.26   P2p


I understand that the network layout here isn't optimial, but I can't 
understand why spanning-tree isn't taking the GE path.

Perhaps I'm misunderstanding how the uplinkfast feature is manipulating 
the costs? I assumed that it would be adding equal costs to the paths that 
C & D see, but perhaps C is acually seeing D's additional 3000 cost, but 
not it's own?

Any help would be very much appreciated. :)

Sam


More information about the cisco-nsp mailing list