When HA (FDM) is disabled, FT virtual machines become unprotected

Products

• VMware vCenter Server

Issue/Introduction

Symptoms:

• When you disable HA (FDM) in vCenter Server 5.0 on a cluster containing Fault Tolerance pairs, the primary virtual machine becomes unprotected without warning
• After you disable HA, fault tolerance virtual machines change to a Not protected state.
  
• In vCenter Server, you see these events:
  
  This virtual machine is not vSphere HA Protected.
  Fault Tolerance state changed from Running to Need Secondary
  
  
• In the vpxd.log file, you see entries similar to:
  
  [ 03800 info 'vpxdvpxdVmomi' opID=179722F2-00000B27-be] [ClientAdapterBase::InvokeOnSoap] Invoke done (w1-vcuiqa069.eng.vmware.com, csi.FdmService.setClusterConfig)
  
  --> dasHostState = (vim.cluster.DasFdmHostState) {
  --> name = "VM002\p",
  --> fullFormattedMessage = "vSphere HA has been disabled in cluster FT of datacenter New Datacenter. vSphere HA will not restart VM VM002 (primary) or its Secondary VM after a failure.",

Environment

VMware vCenter Server 5.0.x

Cause

This issue occurs because FDM disables the FT firewall rule set when HA is disabled.

Resolution

VMware is aware of this issue.

The only workaround when Fault Tolerance (FT) is required to protect virtual machines is to keep HA enabled.

Note: This issue does not occur if there is an NFS datastore mounted to a host. When a NFS datastore is mounted, the NFS client opens a large number of ports, including those used by FT. Therefore, the FT virtual machines remain protected irrespective of whether the FT ruleset is enabled or disabled.

Additional Information

You may want to disable HA when:

• Performing a network maintenance and you do not want HA to take any actions.
• There are HA configuration errors on the host and you want to disable/re-enable HA.

To workaround this issue to avoid disabling HA:

• When performing a network maintenance, disable host monitoring.
  
  When you disable host monitoring, HA detects host failure, but does not respond to the failure.
  
  
• When there are HA configuration errors, select the host list part of the cluster view from the vSphere Client, select all hosts, and then select Reconfigure for vSphere HA.