vSphere High Availability (HA) fails to restart a powered on virtual machine when network partition occurs in VVol environment

Products

• VMware vCenter Server

Issue/Introduction

Symptoms:
When network partition occurs in ESXi mounted VVol datastore, you may experience these symptoms:

• vSphere HA fails to restart virtual machines on VVol datastore protected by HA.
• Virtual machines do not power on as expected.

Cause

This issue may occur when all management network go down and the default gateway of management network is configured as network isolation address for vSphere HA.

Generally, VASA provider uses management network to communicate with vCenter and ESXi. VASA provider cannot guarantee the accessibility to VVol datastores from ESXi if all management network in the ESXi go down.

Therefore, if ‘Shutdown and restart VMs’ or ‘Power off and restart VM’ is configured as the Host isolation response in vSphere HA, it would not be triggered for powered on VM on VVol datastore when the network partition occurs because ESXi host may not access the VVol datastore to power off or shut down and restart VM.

Resolution

VMware provides the Best Practices for Networking for vSphere HA. For more information see vSphere Availability.

Also, ask the VASA vendor about the network redundancy solution of VASA provider for vSphere HA to make sure the supported function of vSphere HA.