Edge BFD tunnels are down after deployment in a stretched Edge cluster

Products

• VMware NSX

Issue/Introduction

• After deploying a new NSX Edge and adding it to an Edge cluster, some BFD tunnels to this Edge are up and some are down.
• The Edge cluster spans two different sites
• vmotion or power off/power on resolves the issue
• After investigation, it is found that only BFD tunnels between Edges in different sites are down

Environment

VMware NSX

Cause

ARP Requests sent by the Edge are dropped in the network between the 2 datacenter sites.
Since the Edge cannot resolve the MAC address of the peer TEP, the BFD tunnel cannot be initialised and remains down.
Vmotion and Power on operation events generate a GARP which can be propagated across sites and resolves the issue.

Resolution

This is not an NSX issue.

Investigate the cause of the failure to propagate ARP Request cross site and engage the network vendor if needed.