NSX Host TEP vmk11 Missing and Edge Cluster Creation Fails in VMware Cloud Foundation

Products

• VMware NSX

Issue/Introduction

In a newly deployed VMware Cloud Foundation (VCF) environment, ESXi hosts are missing the secondary NSX Host Tunnel Endpoint (TEP) interface (vmk11).

Only a single TEP (vmk10) is present on the hosts. Additionally, attempts to create a new NSX Edge Cluster fail due to overlay topology prerequisite failures associated with the host networking configuration.

Environment

VMware Cloud Foundation (VCF) 4.x/5.x
VMware NSX 4.x

Cause

The number of Host Tunnel Endpoints (TEPs) instantiated on an ESXi host is strictly determined by the Uplink Profile applied via the Transport Node Profile (TNP).

If the Uplink Profile uses a "Failover Order" (Active/Standby) teaming policy, or explicitly maps only one uplink to the overlay network, NSX instantiates only a single  TEP (vmk10).

Dual TEPs (vmk10 and vmk11) are instantiated exclusively when the Uplink Profile utilizes a "Load Balance Source MAC" (Active/Active) teaming policy with two active uplinks.

Resolution

 

1. Log in to the NSX Manager UI.

2. Navigate to System > Fabric > Profiles > Uplink Profiles.

3. Locate the Uplink Profile applied to the ESXi host Transport Node Profile (TNP) for the affected cluster.

4. Edit the Uplink Profile and expand the teaming policy configuration.

5. Modify the Teaming Policy to Load Balance Source MAC.

6. Under the Active Uplinks section, ensure that exactly two active uplinks are mapped for the overlay traffic.

7. Save the Uplink Profile.

8. Monitor the host synchronization to verify that the TNP applies the changes and instantiates the secondary TEP (vmk11) on the ESXi hosts.

9. Retry the NSX Edge Cluster creation.

 

Additional Information

Customer Responsibility: Ensure the physical Top of Rack (ToR) switches are configured to support the Active/Active load balancing teaming policy and that the corresponding overlay VLANs are trunked to both physical uplinks.

Note:  In a VMware NSX environment, vmk10, vmk11, and vmk50 are specific VMkernel interfaces automatically created when a host is configured as an NSX Transport Node. They are considered "exclusive" to NSX because they are dedicated to its internal operations and data path rather than general management or storage traffic.

• vmk10 and vmk11 (Tunnel Endpoints - TEPs): These are the default interfaces used for encapsulated overlay traffic (Geneve/VXLAN). They allow the host to participate in the NSX overlay network by providing the IP address used to tunnel traffic between transport nodes. While they are typically named vmk10 and vmk11, their exact naming can sometimes vary depending on what other VMkernel adapters (like VSAN, NFS, or iSCSI) were already present on the host.

• vmk50 (HyperBus Interface): This interface is used for the HyperBus, which facilitates communication between the ESXi host and specialized service VMs or containers (like those in Tanzu/PKS environments). It typically carries traffic on the link-local range 169.254.x.x.