Symptom

*** Environment *** VPC domain is built between two Nexus 9K switches. VPC peer-link has 2 member ports and VPC keep-alive runs on a L3 p2p link between VPC peers using a dedicated VRF. *** Scenario 1 *** 1. Nexus-1 has VPC secondary role and Nexus-2 is primary 2. Nexus-1 is rebooted 3. One or more peer-link member ports remain up on Nexus-2 4. Bridge Assurance Inconsistency and BPDU receive timeout blocking is detected over the vpc peer-link on Nexus-2 5. Traffic loss over VPC VLANs is observed on Nexus-2, since they are placed on a STP Inconsistent state with reason "VPC Peer-link Inconsistent" 6. Traffic resumes until VPC Peer-link Inconsistency is cleared if: 1- VPC peer-link comes up on Nexus-1 and starts sending STP BPDUs or 2- Peer-link goes down on Nexus-2 *** Scenario 2 *** 1. Nexus-1 has VPC primary role and Nexus-2 is secondary 2. Nexus-1 is rebooted 3. One or more peer-link member ports remain up on Nexus-2, so the switch does NOT take Operational Primary role 4. VPC Peer-Keep-alive status on Nexus-2 is "peer is alive" even though no keep-alive packets have been received for more than 5sec 5. Bridge Assurance Inconsistency and BPDU receive timeout blocking is detected over the vpc peer-link on Nexus-2 6. Traffic loss over VPC VLANs is observed on Nexus-2, since they are placed on a STP Inconsistent state with reason "VPC Peer-link Inconsistent" 7. Peer-link goes down on Nexus-2 but VPC Peer-Keep-alive is up, so Nexus-2 suspends all of its VPC port-channels. Below syslog can be seen: %VPC-2-VPC_SUSP_ALL_VPC: Peer-link going down, suspending all vPCs on secondary. If vfc is bound to vPC, then only ethernet vlans of that 8. At this point full outage is seen until Delay-Restore times expires on both Nexus switches and VPC port-channels are removed from suspended state.

Conditions

Observed when VPC peer is rebooted

Workaround

None

Further Problem Description