ESX host does not participate in Spanning Tree

Products

• VMware vSphere ESXi

Issue/Introduction

It is observed that an ESXi host disconnects from vCenter Server and becomes unmanageable immediately after a second vmnic (uplink) is added to the virtual switch backing the management vmkernel adapter (vmk0). The host remains stable only when a single uplink is used.

Cause

This issue can occur because the physical network switch port connected to the newly added ESXi uplink is misconfigured. When the second vmnic is added, the ESXi host hashes management traffic across both active uplinks based on its load balancing policy. If the physical switch port lacks the correct configuration—such as missing management VLAN tagging, a native VLAN mismatch, or a missing Spanning Tree edge mode (PortFast) setting—the physical switch drops the management packets routed through that specific vmnic.

Resolution

To resolve this issue, the physical switch port configuration for the second vmnic must be remediated:

1. Match Physical Port Configuration: Ensure the physical switch port connected to the second vmnic is configured with the exact same VLAN trunking and access parameters as the primary working vmnic.
2. Enable Spanning Tree Edge Mode: Configure the physical ports as edge ports (e.g., spanning-tree portfast or spanning-tree port-type edge) to ensure they transition to a forwarding state immediately when the link is established.
3. Verify VLAN Tagging: Confirm that the management VLAN (e.g., VLAN 203) is correctly tagged on the physical switch trunk port.
4. Validate Teaming: If you are using a Link Aggregation Group (LAG/LACP), ensure the physical switch ports are correctly bundled into a Port-Channel that matches the ESXi load balancing policy.