Intermittent Connectivity on Distributed Portgroups due to Physical Uplink Mismatch

Products

• VMware vSphere ESXi

Issue/Introduction

Virtual Machines (VMs) or VMkernel (VMK) adapters on a newly created Distributed Switch (vDS) portgroup experience intermittent network connectivity or a total loss of access when residing on a specific host. Troubleshooting may reveal:

• MAC addresses are missing from the physical core switch for affected VMs.
• vmkping tests fail to the default gateway from the affected host.
• Connectivity is restored when the VM is vMotioned to a different host in the same cluster.
• CDP or LLDP neighbor information shows discrepancies in port descriptions compared to physical cabling documentation.

Environment

VMware vSphere

Cause

In environments where a vDS uses multiple physical uplinks (vmnics) with load-balancing or teaming policies, one uplink may be correctly configured on the physical switch (e.g., VLAN trunking) while another is misconfigured or physically swapped. Because traffic can pin to a single uplink, the network may appear functional for some sessions but fail for others, creating the illusion of a random or vDS-wide failure.

Resolution

To verify if the issue is isolated to a specific physical uplink, use the esxtop utility to identify which uplink the affected VM or VMK is currently using.

Verifying Uplink Binding with esxtop

1. Log in to the ESXi host via SSH as root.
2. Type esxtop and press Enter.
3. Press n to switch to the Network view.
4. Locate the VM or VMK adapter in the list.
5. Observe the TEAM-PNIC column to identify which physical NIC (vmnicX) the traffic is currently pinned to.
6. To test the alternative uplink, you may temporarily put the identified vmnic into an 'Unused' or 'Down' state at the portgroup level to force the traffic to failover to the other uplink.