vSphere HA restarted a VM when VM Monitoring is disabled

Products

• VMware vCenter Server
• VMware vSphere ESXi

Issue/Introduction

• vSphere Tasks/Events show that a VM was restarted by vSphere HA.
• The ESXi host running the VM did not have a larger HA event (host failure, etc..) impacting all VM's on the host.
• VM Monitoring is disabled.
• FDM logs on ESXi show the last VM poweroff was unclean prior to failing over:

Location of logs on ESXi: /var/log/fdm.log

[YYYY-MM-DDTHH:MM:SS] In(###) Fdm[#######]: [Originator@#### sub=Invt opID=WorkQueue-########] Treating power off as unclean for inaccessible vm /vmfs/volumes/<volume_name>/<VM_Name>/VM_Name.vmx

[YYYY-MM-DDTHH:MM:SS]  Db(###) Fdm[#######]: [Originator@#### sub=Execution opID=host-########] Failing over vm /vmfs/volumes/<volume_name>/<VM_Name>/VM_Name.vmx (isRegistered=true)

• The vmkernel logs on ESXi may show that the VM crashed.

Location of logs on ESXi: /var/log/vmkernel.log

[YYYY-MM-DDTHH:MM:SS] In(###) vmkernel: cpu##:#######)UserDump: ####: vmx: Dumping cartel ####### (from world #######) to file /vmfs/volumes/<volume_name>/<VM_Name>/vmx-zdump.001 ...

Environment

All versions of vSphere

Cause

If a VM is restarted for a reason that is considered an unclean power off then vSphere HA will restart the virtual machine regardless of VM Monitoring settings. Unclean power offs can happen for several reasons including:

1. Powering off a VM in the vSphere UI with the "Hard stop" option. 
2. The VM crashes on ESXi (VMX crash). 

Resolution

It is by design that vSphere HA will restart VM's regardless of VM monitoring settings if the shutoff is unclean. To avoid vSphere HA events on planned VM shutdowns, it is recommended to use a "Power Off" in the vSphere UI.

If the vSphere HA VM restart was due to a VMX crash, contact Broadcom Support for further analysis. 

Additional Information

If VM Monitoring is enabled, see A specific virtual machine was restarted by vSphere HA.