Hello Chris,
Regarding to your question, you just need to have one root bridge in the whole L2 domain so the priorities between switches need to be the same, for example:
switch 1 --> 1 for vlan 10
switch 2 --> 2 for vlan 10
switch 3 --> 3 for vlan 10
switch 4 --> 4 for vlan 10
So in this case the switch 1 will be the root for the whole tree, if you are running the default rstp, if you are using MST that would be based on instances.
As you can see above this is handled in the same way as with regular switches not running vPC, they will use their own MAC to send the vPC, the only difference is that the primary switch in the vPC domain (sw#1 with sw#2, sw#3 with sw#4) is the one in charge of processing the BPDUs no matter if the root is located in the other switch.
For example:
If switch #1 is the root but the primary in the vPC domain is switch #2 ( you can check that with the "show vpc role" command, that election takes place based on the root priority value under the vPC domain, the lower the better), then switch# 2 will be in charge of processing the BDPUs, if switch #1 receives a BPDU, the BPDU is sent accross the peer-link to the other switch to be processed.
So if you keep the setup like this, the priorities need to be different across the switches, to select one of the 4 as the root for the VLAN.
---------
Now, I can see that you have configured the feature "peer-switch" under the vPC domain. When peer-switch is enabled, each Nexus switch shares a virtual bridge ID which allows both switches in the domain to act as root for the VLAN. The vPC peer-link is always in a forwarding status and runs L2 Gateway Interconnection Protocol (L2GIP) in order to prevent bridging loops.
Each Nexus switch sends BPDUs with with the system MAC they both shared form the vPC domain creation, that MAC is 00:23:04:ee + the domain #.
The main advantage of vPC peer-switch is the improvement in term of convergence time during vPC primary peer device failure/recovery. Without vPC peer-switch feature, vPC primary peer device failure and recovery usually create around 3 seconds of traffic disruption (for south to north traffic). With vPC peer-switch, traffic disruption is lowered to sub-second value because peer device down an up events do not generate any Rapid Spanning Tree Protocol Sync behavior (from a STP standpoint, there is no change in L2 topology.
Another enhancement with this feature is that both of the switches will process BPDUs.
The requirements to run properly the peer-switch feature are the following:
1- Configure the peer-switch feature under the vPC domain in both switches (already done in your case)
2- Configure the same priority in both switches, that is because both of the switches will be root, example assuming the 4 switches have peer-switch enabled:
sw#1 and sw#2 --> priority 10 for vlan 10 --> they both will appear as root for that VLAN to the rest of the tree.
sw#3 and sw#4 --> priority 20 for vlan 10 --> sw#4 will see the root in the peer-link and sw#3 will see the root in the connection to SW2.
In summary with the configuration you have (peer-switch enabled), my recommendation would be to follow the last example matching the same priority o a peer-domain basis.
One more thing, technically the only pair of switches that need to have peer-switch enabled are the ones that are going to be the root, so if we put everything together it will look like this:
sw#1 and sw#2 --> peer-switch enabled, priority 10 for vlan 10 --> they both will appear as root for that VLAN to the rest of the tree.
sw#3 --> peer-switch disabled, priority 20 for vlan 10 --> will see the root in the connection going to SW2
sw#4 --> peer-switch disabled, priority 30 for vlan 10 --> will see the root in the connection going to SW3.
More information about peer-switch can be found in the following link:
https://www.cisco.com/c/en/us/support/docs/switches/nexus-7000-series-switches/118858-config-nexus7000-00.html
https://www.cisco.com/c/en/us/td/docs/switches/datacenter/sw/5_x/nx-os/interfaces/configuration/guide/if_cli/if_vPC.html#pgfId-1829433
Let me know if you have questions.
HTH.
Orlando :)
- Orlando -
