MSTP configuration not running

[antilope1]

Hello, 

I have two switches running as root for 2 instances in MSTP, each one is root for a different instance: 

	vlans	root
MST1	6  30	A (2960X)
MST2	2  49  50  168	B (3750G)

and then another switch model 3750G (called Switch) connected to both roots via 2 dot1q trunks, one to each root.  

So, the situation which I expected was to have trunk ports as:

 root port for the vlans in MST1 for the port connected to A 

 root port for the vlans in MST2 for the port connected to B

Switch		Trunks for Switch to Root		vlans  to be root port
Switch	Gi1/0/23	Root A	gi2/0/47	6  30
Switch	Gi1/0/24	Root B	gi1/0/21	2  49  50  168

However, what I find is that both roots find themselves as root for every instance and not Root A root for MST1 and Root B as root for MST2. 

For example, the roots for Root A is always itself:  

Primary#sh spanning-tree root 

                                        Root    Hello Max Fwd
MST Instance           Root ID          Cost    Time  Age Dly  Root Port
---------------- -------------------- --------- ----- --- ---  ------------
MST0              4096 1ce8.5d36.8c80         0    2   20  15                  
MST1              4097 1ce8.5d36.8c80         0    2   20  15                  
MST2             28674 1ce8.5d36.8c80         0    2   20  15

And for Root B, it finds itself for MST1-2 but not for MST0: 

Secondary#sh spanning-tree  root

                                        Root    Hello Max Fwd
MST Instance           Root ID          Cost    Time  Age Dly  Root Port
---------------- -------------------- --------- ----- --- ---  ------------
MST0              4096 1ce8.5d36.8c80     20000    2   20  15  Gi1/0/22        
MST1             28673 001b.2ad9.8300         0    2   20  15                  
MST2             24578 001b.2ad9.8300         0    2   20  15

Is this normal? Why both roots don't recognize themselves in MST1-2? 

I attach spanning-tree summary for both roots and configs for 3 switches (A y B, and Switch which is not root). 

Thank you very much in advance. Please, any comments or questions will be grateful. 

[Peter Paluch · Accepted solution]

Hi,

Your switches are not in the same MSTP region. Your SwitchA uses revision 1 while the other switches lack this command and use revision 0 as a result (this is confirmed by their output of the show spanning-tree mst configuration command). For switches to be in the same MST region, they must match in all configurable parameters, that is, region name, revision number, and VLAN-to-instance mappings.

To solve your problem, either add the revision 1 command to your remaining two switches, or remove this command from SwitchA's configuration.

Best regards,
Peter

[antilope1]

Hi Peter, 

Yes, I changed revision number to 1 and now it makes sense. 

Now I see everything running ok. I suppose when combining Cisco with HP is better to go to MSTP than stay with rapid-pvst and the somehow compatible protocol of HP. This is the reason to try MSTP. 

Any advise in migration from rapid-pvst to MSTP? 

Thank you very much!! 

[Peter Paluch]

Hi,

You are welcome.

When combining switched from multiple vendors, MSTP is the proper way to go even though maintaining it may be a little more cumbersome.

When migrating from Rapid-PVST to MSTP, well, there have been quite a few thread on that here already. The key takeaways:

• The migration should be performed during a maintenance window. There will be transient connectivity outages.
• Before migrating any switch to MSTP, make sure that all are running RSTP and that all switches have their edge ports properly configured as edge ports (this is PortFast on Cisco switches)
• Make sure that all switches in all VLANs currently have their priorities set to a value higher than 8192. The reason for this is that during the migration, the MSTP region will need to be configured as a "root switch" for all VLANs in the Rapid-PVST region. For this, it is necessary to have some "free space" in the switch priorities so that there is still a place to go down with.
• Apply the MST region configuration to all switches in your network but leave them running in Rapid-PVST mode. The MST region configuration will be inactive until you start MSTP on the switch but it will be prepared to kick into action.
• Starting with a single-region design, decide which switch is going to be the root of the IST (MSTI 0). Configure this switch with the MSTI 0 priority of 8192 or 4096; in general, to any value that is lower than the priority of any Rapid-PVST switch in any VLAN. Then convert this switch to MST. Because of the way MST and PVST interoperation works, the priority of the IST root switch will be replicated into all per-VLAN instances on the MST region boundary, causing the Rapid-PVST switches to consider the entire MST region to be their "root switch", and this is what you want.
• Now gradually convert the other switches to MST, progressing one layer ("hop") further from the already converted switches at a time, always verifying that the connectivity is maintained. Because Cisco switches do not implement MST/Rapid-PVST interoperability, only MST/PVST, in combined MST and Rapid-PVST regions, the STP operations on the region boundaries will be downgraded to legacy STP, causing outages for up to 50 seconds (though hopefully shorter).

Best regards,
Peter

[antilope1]

Hi Peter, you have been very kind. Thank you for your long reply.... 

I am very grateful. The best for you!