[c-nsp] Solution to %SPANTREE-2-RECV_PVID_ERR, except disable spanning tree?

[Clinton Work]

Cisco PVST+ / RPVST do integrate the vlan ID into the bridge priority 
(bit stealing), but that is a function of MAC reduction to support 4096 
VLANs rather than PVST+ proper.  MAC reduction will do the same thing 
with regular 802.1d BPDU priority values and you can interconnection two 
Vlans with 802.1D without causing problems.  The PVST+/RPVST BPDU format 
has an extra TLV added onto the end with an attribute called PVID.  The 
PVID attribute is simply the vlan that the STP BPDU is intended for.  
All the Spanning-Tree PVID error detection and messages are based upon 
the PVID TLV. 

Clinton. 

Benjamin.Conconi at nok.ch wrote:
> We had a similar problem a time ago. We did some tests with a cisco es20
> linecard and eompls services. This card has a feature called
> vlan-translation were you can translate one vlan to a other. So we had a
> setup like this
>
>
> |-----|            |-------|            |-----|
> |2960 |--Vlan 2412-|Eompls |--Vlan 2413-|2960 |
> |-----|            |-------|            |-----|
>
> The problem is, that the PVSTP couples the vlan id with the bridge
> priority. Means if you let your bridge priority by default, the bridge
> priority of vlan 2412 is 32768 + 2412 = 35180. Hence the bridge priority
> of vlan 2413 is 35181. You can verify this with the command sh
> spanning-tree.
>
> Both, the bridge priority and the vlan id are integrated in the bpdu of
> the pvstp algorithm. If the left-hand switch sends a bpdu to the eompls
> core, the vlan id 2412 will be translated to 2413, but the bridge
> priority remains. Because the bridge priority and the vlan id are
> coupled by a simple addition, the right-hand switch (2960) detects, that
> something changed the vlan id. The Switch will then block the vlan and
> it apperas with the message you got:
>
>   
>


-- 

=========================================================
Clinton Work
Airdrie, AB