Inconsistent BPDU Forwarding Between vPC Peers After Switch Reboot

[nunzio-cordalonga]

Hi all,

I’m facing a strange behavior in a network topology with two pairs of Nexus C3548P-10GX switches connected in a fully meshed back-to-back vPC design.

The root bridge is located in the first pair (note: peer-switch is not enabled). It seems that the switch currently acting as root bridge is not sending BPDUs on the vPC interfaces.
On the other hand, the second switch of the pair — which became operational primary after a recent reboot — is sending BPDUs on all vPCs, even on one that is currently down.

Additionally, we discovered that the physical cabling between the two vPC domains was not correctly implemented to form a fully meshed topology.
They mistakenly connected two vPC interfaces from NX1 directly to LNZ1, and similarly from NX2 to LNZ2, without cross-connecting (i.e., no NX1→LNZ2 or NX2→LNZ1). So the interconnection is not truly fully meshed as intended.

The strange part is that when we run show spanning-tree on the vPC interfaces on NX1, the BPDU sent counter is stuck at 0.

However, on the other side, LNZ1 shows an increasing BPDU received counter, as if BPDUs are still being received — despite NX1 not reporting any being sent.

The two Nexus pairs are connected via vPC, and this inconsistent BPDU behavior is creating confusion in the spanning tree topology.

I’d be happy to share show command outputs (e.g. show vpc, show spanning-tree, interface details) or a network diagram if it helps clarify the situation.

Has anyone encountered a similar issue? Could the incorrect cabling or role mismatch be the root cause?
Any advice would be greatly appreciated.

[MHM Cisco World]

Inconsistent BPDU is only happened when you run in PVST in one pairs and MST in other pairs 

MHM

[nunzio-cordalonga]

I can confirm that all four Nexus switches are running RSTP.
So the inconsistency doesn’t seem to be related to mismatched STP modes between the vPC domains.

[MHM Cisco World]

Then check 

Trunk between two pairs allow specific vlan' 

Check this vlan is available in both pairs or not' 

Focus on vlan appear in Inconsistent bpdu log 

MHM

[Enes Simnica]

hello man. ur situation looks like a classic case of STP instability due to vPC misconfiguration and improper cabling.

The root cause is the lack of cross-connections between the two vPC domains (NX1-LNZ2 and NX2-LNZ1). Without these, the topology isn’t fully meshed, which breaks redundancy and prevents BPDUs from propagating correctly. This asymmetry is likely why one side is seeing BPDUs while the other shows none being sent.

Another key issue is that peer switch is not enabled. Without it, only the vPC primary sends BPDUs, and if the STP root happens to be the secondary, it won't send any  leading to the confusion u're seeing now. Enabling peer-switch allows both vPC peers to act as a single logical bridge in STP, which is crucial for stability.

if u ask me I recommend starting by fixing the cabling to establish a proper full mesh between the vPC domains. Then, enable spanning-tree vpc peer-switch on both vPC pairs.

Once that’s done, run the well known: show spanning-tree vpc consistency, show vpc consistency-parameters, show vpc 

If the BPDU issue still persists after those changes, it would help to see:show tech-support spanning-tree, sh vpc statistics, and sh stp timer settings...

hope it helps G

-Enes

[nunzio-cordalonga]

That definitely sounds like the right direction to me as well.
Fixing the cabling and enabling vpc peer-switch on both vPC domains should help bring stability and eliminate the BPDU inconsistency.

I was also considering upgrading all four Nexus switches, since they're currently running on NX-OS 9.2(1) from 2018. I plan to bring them up to the suggested release 9.3(13) from 2023 — that might resolve additional bugs or quirks as well.

I’ll definitely try all this in the coming weeks. Fingers crossed!

[Enes Simnica]

I’ll have some free time tomorrow, and I’ll simulate ur topology on EVE-NG. Let’s see what we get from it........

[julian.bendix]

Is there a specific reason why peer-switch is not enabled?

[nunzio-cordalonga]

Honestly, I’m not sure — this is a customer’s network that I had the chance to analyze, and even they don’t really know the reasoning behind these design choices.
Apparently, the person who previously managed the network left without providing any documentation or explanations.

[julian.bendix]

Read through the peer-switch Section of the following document while reviewing your case, then consider enabling it  
https://www.cisco.com/c/en/us/support/docs/ios-nx-os-software/nx-os-software/217274-understand-virtual-port-channel-vpc-en.html

After the cabling issues are solved of course!