VMs intermittently lose network connections after vMotion

Products

• VMware vSphere ESXi

Issue/Introduction

A working VM loses network connectivity when vMotioned to a new host. Connectivity can be recovered by disconnecting and reconnecting the VM NIC but, this does not always work. This can happen to multiple VMs, but other VMs on the host and on the same VM network are working. In some cases, a VM went down after migrating to the newly configured host.

 

Environment

VMware vSphere (all versions)

Cause

If the VM went down after migrating to a newly configured ESXi host, first check which physical uplink NIC it was using.

Running ESXtop reveals that all of the VMs that cannot connect are on the same vmnic. The vmnic is selected by an internal process and can change when the NIC is disconnected and reconnected or if vMotioned to another host then, moved back to the troubled host.

Moving that vmnic into "unused" on the DVS resolves the issue but, leaves the DVS without redundant connections.

This is caused by an issue with the NIC and/or chassis. There could also be a network misconfiguration in the upstream network.

Resolution

Contact the internal networking team and have them check the upstream physical device configuration. This could include the server chassis as well as the physical top of rack network switch/router.

Contact the hardware vendors and have them test the hardware and configuration.