VM connectivity fails across ESXi hosts on same subnet due to NIC teaming mismatch - VMware vSphere ESXi

Products

• VMware vSphere ESXi

Issue/Introduction

Virtual machines (VMs) on the same subnet but residing on different ESXi hosts cannot establish bidirectional communication. While ARP resolution and ICMP Echo Requests are successful, the return traffic (ICMP Echo Reply) is dropped before reaching the source host.

Environment

Vmware vsphere Esx

Cause

The issue occurs due to a configuration mismatch between the ESXi virtual switch and the physical switch.

When the physical switch is configured for a Static EtherChannel, the ESXi host must use the "Route based on IP hash" load balancing policy. Refer kb: Vswitch configuration with IP Hash

Using "Route based on source MAC hash" with a Static EtherChannel causes the physical switch to drop return packets or forward them to the incorrect port.

Resolution

To resolve this issue, align the ESXi NIC teaming policy with the physical switch configuration:

1. Log in to the vSphere Client.
2. Navigate to the ESXi Host > Configure > Networking > Virtual switches.
3. Select the affected vSwitch or Distributed Port Group and click Edit.
4. Navigate to Teaming and failover.
5. Change the Load balancing policy to Route based on IP hash.
6. Ensure all physical uplinks (e.g., vmnic1, vmnic5) are in the Active adapters list.
7. Click OK to save the changes.
8. Verify that bidirectional ICMP traffic is restored.

Additional Information

or detailed packet capture instructions, refer to Using pktcap-uw in Esx