TEP BFD Tunnels Dynamically Flapping (Down/Init/Up) on ESXi Hosts.

Products

• VMware NSX

Issue/Introduction

• In an NSX environment, administrators may observe TEP BFD tunnels frequently flapping between down, init, and up states within a short span of time.
• Review of the vmkernel.log on affected ESXi hosts shows messages similar to the following:
  

  2025-10-16T00:04:44.539Z In(182) vmkernel: BFDDeleteSessionEntryLocked:642:[nsx@6876 comp="nsx-esx" subcomp="bfd"]local: ##.##.182.7, remote: ##.##.183.8, type: overlay
  2025-10-16T00:05:40.524Z In(182) vmkernel: BFD_CreateSession:441:[nsx@6876 comp="nsx-esx" subcomp="bfd"]session: 0xc52867d3, local: ##.##.182.8, remote: ##.##.183.8, type: overlay

• No actual datapath impact or network connectivity issues are reported for active workloads.
• Tunnel state changes often occur in approximately two-hour intervals or correlate with VM lifecycle events.

Environment

VMware NSX

Cause

This behavior is an expected operational characteristic of NSX dynamic tunnel management.

TEP tunnels are dynamically established and torn down based on the presence of overlay-connected VMs:

1. Dynamic Deletion: When the last VM utilizing a specific overlay network is powered off, deleted, or migrated away from a host, NSX deletes the associated BFD sessions and TEP tunnels to conserve system resources.
2. Dynamic Creation: When a VM requiring overlay connectivity is powered on or deployed on a host, the management plane triggers the creation of the necessary TEP tunnels and BFD sessions.

Environments with frequent VM churn (such as Kubernetes/Tanzu clusters, VDI environments, or automated testing frameworks) will see this flapping more frequently as a result of normal VM lifecycle operations.

Resolution

No action is required as this is by design. The flapping indicates the system is correctly managing resources by only maintaining tunnels that are actively required for existing workloads.