MAC Address Learning Issues on Physical Infrastructure for vSphere Distributed Switch (vDS) Connected VLANs

Products

• VMware vSphere ESXi

Issue/Introduction

 

• Virtual machines (VMs) connected to a specific vSphere Distributed Switch (vDS) port group or VLAN experience total network connectivity loss.

• MAC addresses of the affected VMs are not populating in the upstream physical switch MAC address table.

• This issue most commonly occurs during the provisioning of new VLANs or post-maintenance network reconfigurations.

• Environment specific: The hardware fabric utilizes an HPE Synergy blade enclosure architecture with integrated interconnect/fabric modules handling upstream physical infrastructure connectivity.

 

 

Environment

VMware ESX

Cause

This issue is typically caused by a Layer 2 configuration mismatch or omission along the physical network path. Common root causes include:

1. Upstream Trunk Misconfiguration: The newly provisioned VLAN is missing from the allowed/trunk list on the upstream physical switch ports.

2. Fabric/Enclosure Misconfiguration: On HPE Synergy environments, the new VLAN profile or network definition has not been properly mapped or allowed within the HPE OneView Logical Interconnect / Virtual Connect configuration for the specific server profile downlinks.

Resolution

1. Verify vSphere Network Configuration

• Confirm that the affected vDS port group is tagged with the correct VLAN ID.

• Review the Teaming and Failover policy of the port group. Ensure that the designated active uplinks (vmnicX) are operational and showing an UP link status.

2. Perform Packet Capture on the ESXi Host

To determine if traffic is successfully leaving the virtual environment, execute a packet capture on the ESXi host's physical uplink using the CLI.

Log in to the ESXi host via SSH as root and run the following command (replace vmnic#### with your actual uplink name and XX:XX:XX:XX:XX:XX with the VM's MAC address):

pktcap-uw --uplink vmnic#### --capture UplinkSndKernel,UplinkRcvKernel -o - | tcpdump-uw -enr - | grep -i "VM-MAC-ADDRESS"

Analyze the Capture Results:

• Scenario A: Packets Leave (Snd) but Do Not Return (Rcv) If you observe outbound packets leaving the host but no inbound reply traffic, the vSphere configuration is working correctly. The packet is being dropped by the physical infrastructure, or return traffic is blocked.

• Scenario B: No Outbound Packets (Snd) Captured If no packets are seen leaving the uplink, inspect the VM's internal guest OS network configuration, verify that the OS network stack is healthy, and ensure that the vDS port is not blocked or experiencing a local security policy restriction (e.g., MAC Address Changes/Forged Transmits policy).

Validate Fabric and Physical Infrastructure Settings

If Scenario A is confirmed (packets leave the ESXi uplink but the physical switch fails to learn the MAC address):

• For HPE Synergy Frameworks: Log into HPE OneView and verify the Logical Interconnect Group (LIG) and Uplink Set configurations. Ensure the specific VLAN is explicitly defined and associated with the network set applied to the server blade profile.

• For Physical Switches: Verify that the upstream switchport interfaces connected to the enclosure are configured as trunk ports and that the new VLAN is added to the active allowed VLAN list.