Solved: Hello,Interworking MST and

Mokhalil82 · ‎07-09-2015

Hi

I have 2 main sites connected using a layer 2 National Ethernet link. The core switches at each site connected into a smaller 8 port switch which then connects to the circuit as follows:

Core site 1 ---------------8Port SW--------------NE Circuit--------------------------------------------NE Circuit------------8portSW---------------Core site 2

One site is using MST and the other is using Rapid PVST. I have just installed the 2 8port switches at each end and the link between the sites is active as i can see the other site via cdp. But I cannot ping the other site because spanning tree is blocking the link.

I am only running 3 of the many vlans over this link, is there a vlan I can get spanning tree at each end to not block the 3 vlans.

Spanning tree output at site one is below, so its G0/10 that needs to allow the 3 vlans

Thanks

MST0
Spanning tree enabled protocol mstp
Root ID Priority 8193
Address 0008.e3ff.fc28
Cost 20003
Port 10 (GigabitEthernet0/10)
Hello Time 2 sec Max Age 20 sec Forward Delay 15 sec

Bridge ID Priority 32768 (priority 32768 sys-id-ext 0)
Address f078.16f6.fb80
Hello Time 2 sec Max Age 20 sec Forward Delay 15 sec

Interface Role Sts Cost Prio.Nbr Type
------------------- ---- --- --------- -------- --------------------------------
Gi0/1 Desg FWD 20000 128.1 P2p
Gi0/2 Desg FWD 20000 128.2 P2p
Gi0/10 Root BKN*20000 128.10 P2p Bound(PVST) *PVST_Inc

MST1
Spanning tree enabled protocol mstp
Root ID Priority 24577
Address e840.40fe.5200
Cost 20000
Port 1 (GigabitEthernet0/1)
Hello Time 2 sec Max Age 20 sec Forward Delay 15 sec

Bridge ID Priority 24577 (priority 24576 sys-id-ext 1)
Address f078.16f6.fb80
Hello Time 2 sec Max Age 20 sec Forward Delay 15 sec

Interface Role Sts Cost Prio.Nbr Type
------------------- ---- --- --------- -------- --------------------------------
Gi0/1 Root FWD 20000 128.1 P2p
Gi0/2 Altn BLK 20000 128.2 P2p
Gi0/10 Mstr BKN*20000 128.10 P2p Bound(PVST) *PVST_Inc

MST2
Spanning tree enabled protocol mstp
Root ID Priority 24578
Address 08d0.9f57.e080
Cost 20000
Port 2 (GigabitEthernet0/2)
Hello Time 2 sec Max Age 20 sec Forward Delay 15 sec

Bridge ID Priority 28674 (priority 28672 sys-id-ext 2)
Address f078.16f6.fb80
Hello Time 2 sec Max Age 20 sec Forward Delay 15 sec

Interface Role Sts Cost Prio.Nbr Type
------------------- ---- --- --------- -------- --------------------------------
Gi0/1 Altn BLK 20000 128.1 P2p
Gi0/2 Root FWD 20000 128.2 P2p
Gi0/10 Mstr BKN*20000 128.10 P2p Bound(PVST) *PVST_Inc

Peter Paluch · ‎07-09-2015

Hello,

Interworking MST and Cisco PVST is somewhat difficult. You are experiencing a so-called PVST Simulation Inconsistency. The task of PVST Simulation is to make sure that even though only MSTI 0 in MST and VLAN 1 in PVST are allowed to talk to each other, a consistent port role and state determination will be made by all VLANs in the PVST region. This can generally be accomplished only in two ways:

Either configure your MST switch so that its priority in MSTI 0 is lower than the priority of any PVST switch in any VLAN in the PVST region. This will make the MST switch to become the root switch for MSTI0 as well as a root switch for each and every VLAN in the PVST region.
Or configure the PVST switch so that it becomes the root for both the PVST and MST regions:
- In VLAN 1, it has a priority lower than the priority of the current MSTI 0 root
- In VLANs 2-4094, it has a priority lower than its own priority in VLAN 1

The first option is generally preferred, as it is configurationally easier and more easily handled. Ideally, though, you should convert your network to run full MST. Even though MST internally runs RSTP for each instance, on boundary ports between MST and Rapid PVST regions, Cisco switches fall back to legacy STP, thus losing the rapid convergence properties on such links.

Best regards,
Peter

View solution in original post

Peter Paluch · ‎07-10-2015

Hi,

I understand.

Okay, what you can do is truly create two MST regions ideally distinguished by their name. You can keep the instance numbers the same in both regions if it suits you, or you can use separate instance numbers. Also, the mapping of VLANs onto instances is entirely at your discretion.

By having multiple regions, the only instance that spans across all regions is the instance 0. On boundary links between regions, this is the only instance that speaks on behalf of the whole region to the other region. Changes to instance 0 inside a region can therefore propagate throughout the whole network. Therefore, it is best to move all VLANs out instance 0 in all regions so that the impact of reconvergence in instance 0 is minimized on them.

There is, in general, no need to create a standalone MST instance for the 3 VLANs that span both regions, and map these VLANs into this instance identically in both regions. Even if you did that, the regions would still treat each other as a separate entity (which they truly are), and the only instance they would use to speak to each other in MST would be instance 0. The role and state of boundary ports is determined by the two instances 0 from both regions talking to each other, and all other instances and thus all VLANs merely follow these roles and states. The fact that you have the same instance in both regions with the same VLANs mapped onto it would make absolutely no difference in such case.

I take it if same instances at each site have different vlans they will participate in their own stp election ie mst 1 in site 1 has different vlans to mst 1 in site 2 so their stp election is different?

Yes, these two MST instances number 1 would be independent, and the spanning tree calculation for them would run independently in each region.

Would it be worth using instance 3 at both sites that only contains the spanned vlans or will mst1 and 2 at both sites with different vlans work.

It would make no difference to create instance 3 for the spanning VLANs, apart from the possibility of having a separate root and thus a separate spanning tree for this instance 3 in each region. Otherwise, though, there is no advantage in doing that.

What I am trying to avoid is that any changes in mst 1 at site 1 to cause reconvergence across both sites to mst 1.

Having multiple regions will accomplish this goal.

Also i would like the core switch at site 2 to be the root for its vlans and site 1 core to remain the root for its vlans.

Having multiple regions and mapping VLANs into instances other than instance 0 will accomplish this goal.

Best regards,
Peter

View solution in original post

Reza Sharifi · ‎07-09-2015

Hi,

MST is backward compatible with PVST. If these are new sites you are bringing up and not in production yet, can you try to run the same type of STP?

What type of switches are these?

Can you post STP configs

HTH

Peter Paluch · ‎07-09-2015

Hello,

Interworking MST and Cisco PVST is somewhat difficult. You are experiencing a so-called PVST Simulation Inconsistency. The task of PVST Simulation is to make sure that even though only MSTI 0 in MST and VLAN 1 in PVST are allowed to talk to each other, a consistent port role and state determination will be made by all VLANs in the PVST region. This can generally be accomplished only in two ways:

Either configure your MST switch so that its priority in MSTI 0 is lower than the priority of any PVST switch in any VLAN in the PVST region. This will make the MST switch to become the root switch for MSTI0 as well as a root switch for each and every VLAN in the PVST region.
Or configure the PVST switch so that it becomes the root for both the PVST and MST regions:
- In VLAN 1, it has a priority lower than the priority of the current MSTI 0 root
- In VLANs 2-4094, it has a priority lower than its own priority in VLAN 1

The first option is generally preferred, as it is configurationally easier and more easily handled. Ideally, though, you should convert your network to run full MST. Even though MST internally runs RSTP for each instance, on boundary ports between MST and Rapid PVST regions, Cisco switches fall back to legacy STP, thus losing the rapid convergence properties on such links.

Best regards,
Peter

Mokhalil82 · ‎07-10-2015

Thanks for the comments

So the good thing I suppose is that site 2 is a new site and I set up rpvst as i was more familiar with it. Its not in productions so I can change it to MST.

But both sites have different vlans and different subnets and we want to keep it like that, apart from only allowing 3 vlans over between them which will carry some replication traffic.

So how would i setup MST at site 2, can I have it running a separate mst region but then both regions allow these 3 vlans to run between them?

Or if I keep RPVST, i take it if the root bridge in at site 1 has a priority of 27xxx I need to configure the root bridge at site 2 to have a priority of higher number so lower priority, this will cause the 3 vlans running between the sites to use the MST bridge as its root and then not put that port in BKN status?

Thanks

Peter Paluch · ‎07-10-2015

Hello,

So how would i setup MST at site 2, can I have it running a separate mst region but then both regions allow these 3 vlans to run between them?

Don't worry, the scope of VLANs is not limited to MST regions. Your 3 VLANs will cross the MST boundaries just fine.

Personally, just for the sake of simplicity, I would suggest keeping a single MST region and mapping all VLANs (both existent and non-existent) to a MST Instance 1. Leave the MST Instance 0 unused. Simply put, this would be the MST configuration on all of your switches:

spanning-tree mode mst
spanning-tree portfast default ! Important in RSTP/MSTP networks
spanning-tree mst configuration
 name MyRegion
 revision 1
 instance 1 vlan 1-4094

Even though your two sites have in general different VLANs, they will still form a single tree in instance 1 that will be shared by all VLANs that happen to exist.

Does this sound okay to you?

Best regards,
Peter

Mokhalil82 · ‎07-10-2015

Hi Peter

The only issue with the recommendation is that site 1 is an existing network which has 2 mst instances, one for odd and one for even vlans and i wont have authority to change those.

So with that said I would like to do the same at site 2 and have 2 instances. I take it if same instances at each site have different vlans they will participate in their own stp election ie mst 1 in site 1 has different vlans to mst 1 in site 2 so their stp election is different?

Would it be worth using instance 3 at both sites that only contains the spanned vlans or will mst1 and 2 at both sites with different vlans work.

What I am trying to avoid is that any changes in mst 1 at site 1 to cause reconvergence across both sites to mst 1. Also i would like the core switch at site 2 to be the root for its vlans and site 1 core to remain the root for its vlans. Reason i said instance 3 for the spanned vlans is to allow me to do this in hours as we cannot afford downtime due to reconvergence at site 1. So if mst 3 is used for the spanned vlans then i can make changes as i wish and it would only reconverge mst3.

Thanks

Peter Paluch · ‎07-10-2015