Virtual Machines fail to restart via vSphere HA during All Paths Down (APD) events.

Products

• VMware vCenter Server

Issue/Introduction

Following an All Paths Down (APD) storage event, a significant number of Virtual Machines (VMs) fail to be restarted by vSphere High Availability (HA) despite the cluster being configured for the same.

Symptoms:

• During a storage APD event, only a subset of impacted VMs are successfully failed over or restarted by HA.

• vSphere HA (FDM) logs contain the entry: Both APD terminate and reset are disabled for VM <VM_PATH>. Ignore the failure..
  
  2026-01-17T20:25:08.459Z In(166) Fdm[#####]: [Originator@6876 sub=Policy opID=WorkQueue-######] Both APD terminate and reset are disabled for VM /vmfs/volumes/#########-#########/VM_NAME/VM_NAME.vmx. Ignore the failure.
2026-01-17T20:25:08.462Z In(166) Fdm[#####]: [Originator@6876 sub=Policy opID=WorkQueue-######] Both APD terminate and reset are disabled for VM /vmfs/volumes/#########-#########/VM_NAME_1/VM_NAME_1.vmx. Ignore the failure.
.
.
2026-01-17T20:26:28.700Z In(166) Fdm[#####]: [Originator@6876 sub=Policy opID=WorkQueue-######] Both APD terminate and reset are disabled for VM /vmfs/volumes/#########-#########/VM_NAME_2/VM_NAME_2.vmx. Ignore the failure.
2026-01-17T20:26:28.702Z In(166) Fdm[#####]: [Originator@6876 sub=Policy opID=WorkQueue-######] Both APD terminate and reset are disabled for VM /vmfs/volumes/#########-#########/VM_NAME_3/VM_NAME_3.vmx. Ignore the failure.

• The storage impact duration exceeds the configured HA response delay, but no action is taken for specific VMs.

Environment

vSphere vCenter 7.x

vSphere vCenter 8.x

 

Cause

The issue is caused by VM Overrides configured at the cluster level. When a VM has an override specifically setting the APD Protection (Response for All Paths Down) to Disabled, vSphere HA is explicitly instructed to ignore storage connectivity failures for that specific object.

In this state:

• HA will not "terminate" (kill) the VM process on the impacted host.

• Consequently, HA cannot restart the VM on a healthy host because the original VM process is still registered and "running" (albeit unresponsive) on the host experiencing the APD.

• The following setting for the APD Failure Response (Disabled) can be seen on the VM Overrides tab.
  
  

Resolution

To ensure all VMs restart during an APD event, the VM-specific overrides must be removed or modified to follow the cluster default settings.

1. Log in to the vSphere Client.

2. Navigate to the Hosts and Clusters view and select the affected Cluster.

3. Click the Configure tab.

4. Under Configuration, select VM Overrides.

5. Identify the VMs listed with APD Protection set to Disabled.

6. To restore default HA behavior:

   • Select the affected VM(s) and click Delete to remove the override entirely.

   • Alternatively, click Edit and change the Failure Response under the APD Protection tab to Use Cluster Settings or a specific restart policy (e.g., Power off and restart VMs).

Additional Information

Customize an Individual Virtual Machine