VM power-on fails with "VM is pinned to a host" error despite no active affinity rules

Products

• VMware vSphere ESXi
• VMware vCenter Server
• VMware Telco Cloud Platform
• VMware Cloud Director

Issue/Introduction

• Powering on a virtual machine (VM) in a vSphere cluster fails with the error: Unable to power on the VM, the status shows as VM is pinned to a host.
  
  
• The vCenter Server logs (vpxd.log) :
  

  [VpxLRO] -- ERROR task-111113 --  -- <VM_name> -- Datacenter.ExecuteVmPowerOnLRO: :vim.fault.DrsDisabledOnVm
  --> Result:
  --> (vim.fault.DrsDisabledOnVm) {
  -->    faultCause = (vmodl.MethodFault) null,
  -->    faultMessage = <unset>
  -->    msg = ""
  --> }
  --> Args:
  -->
  

   

• DRS is enabled on the cluster and set to Fully Automated.
  
  
• No "Must run on hosts in group" affinity rules or VM-VM anti-affinity rules are visible in the vSphere Client that would explain the pinning.
  
  
• Disabling anti-affinity rules (if any exist) or manually migrating the VM to another host allows it to power on.

Environment

VCD 10.4.3

Vcenter 8.0.3

ESXi 8.0.3

TCP 5.0.1

Cause

• This issue is caused by stale entries in the vCenter Server database or memory that incorrectly maintain a "Hard Rule" (strict pinning) constraint for the VM.
• This typically occurs after a placement policy change or a failed automated deployment (such as via VMware Cloud Director or Nokia CBAM) where the LRO (Long Running Operation) API fails to clear the initial placement constraints.

Resolution

To resolve this issue, perform a restart of the vCenter Server services to flush the stale metadata.

1. Log in to the vCenter Server Appliance Management Interface (VAMI) at https://vcenter_fqdn:5480.
   
   
2. Navigate to Services.
   
   
3. Select the VMware vCenter Server service (vpxd) and click Restart.
   • Note: This will temporarily interrupt vCenter management but will not affect running VMs.
     
     
4. Once the service is back online, attempt to power on the VM again.

Workaround:

If a service restart cannot be performed immediately:

1. Temporarily disable any VM/Host Affinity Rules for the affected cluster.
2. Manually migrate the VM to a host with known available capacity.
3. Power on the VM.
4. Re-enable the rules; DRS should now correctly track the VM's placement without the stale "pinned" state.

If it is VCD VM, try to disable below option in Anti-affinity rule configuration and then try powering ON: