MST and PVST

[LucaSalvatore_2]

Is it possible to have a switch, outside of an MST region, running PVST that is the root switch for some VLANs?

We have a L2 link to an ISP, the ISP has set their switch to STP priority 0 for all VLANs, so it is the root switch of our network and is creating some sub optimal L2 performance.  They won't change the priority of their switch so I need a new solution. 

I was thinking i could setup an MST region for all my equipment and leave their equipment out of it.  But when I set it up, on the link that connects to the ISP i get these errors : SPANTREE-2-PVSTSIM_FAIL: Superior PVST BPDU received on VLAN.

Any ideas?  Can I somehow make my switch the root for my VLANs when their switch is configured with priority 0.

[smitesh kharecha]

Hi Luca,

Remember that Root ID in STP (any flavour) is priority + mac address.

So if the priority is 0, the only thing which can help you to make your device ROOT is lower mac address.

So, basically, you would also set priority 0 and will select root with lower mac address.

There is one more thing which you can try (as I'm not sure , whether it will work) is, try the following command on the switch you want to make root:

spanning-tree vlan (your vlan id) root primary

HTH,

Smitesh

[Peter Paluch]

Hi Luca,

The interoperation of MST and PVST is problematic, because it is not possible to represent per-VLAN STP state as utilized by PVST inside a MST region where per-VLAN semantics does not exist. For example, how would you deal with a situation when two VLANs have different root bridges in the PVST region but map to the same MST instance?

In pure IEEE environment, the situation is simple: there is either MST with its instances, or (R)STP with no instance or VLAN knowledge. The entire network therefore consists of MST regions and the remaining (R)STP region, if any. Either one of the MST regions becomes the root bridge in the entire switched network, or the (R)STP region will. There is no other option, nor is it necessary.

In PVST, things get much more complicated. The MST/PVST interoperation tries to mimic the IEEE approach by also requiring that either an MST region becomes the root bridge for all VLANs, or the PVST region does. However, what complicates things is the fact that in the PVST region, there are as many STP instances as there are VLANs, and their settings may be incongruent: some VLANs may have the root bridge in the PVST region, some others may point back to the MST region for the root bridge. So Cisco really allows only two possible configurations:

• Either the MST region becomes the root bridge for all VLANs. This is accomplished by configuring a bridge inside the MST region for the MST instance 0 so that its priority in instance 0 is the lowest among all priorities in the PVST region, on all PVST switches for all VLANs.
• Or the MST region does not become the root bridge for any VLAN. This is harder to accomplish: not only must the priority of any switch in the MST region for instance 0 be higher than any PVST switch's priority for VLAN1 in particular, but also, PVST priorities for VLANs other that VLAN1 must be lower than the priority of the PVST bridge for VLAN1 (it is a sanity check basically saying "I must be sure that if VLAN1 beats me which is the only VLAN I really interact with, I must be sure that other VLANs would beat me as well")

Therefore, it is not possible to accomplish what you are trying to achieve. Either the MST region is the root for all VLANs, or is not a root for any VLAN. Any other combination won't work.

Best regards,

Peter