[VCL-974] VMware VMs may fail to power on if host does not support nested VMs

• Type: Bug
• Status: Resolved
• Priority: Major
• Resolution: Fixed
• Affects Version/s: 2.4.2
• Fix Version/s: 2.5
• Component/s: vcld (backend)
• Assignee: Andrew Kurth
• Reporter: Andrew Kurth
• Created: 21/Jul/16 17:55
• Updated: 28/Apr/17 16:59
• Resolved: 28/Apr/17 16:59

Description

VMware.pm is adding the following to the .vmx file to allow a VM to run nested VMs:

"monitor.virtual_mmu" => "hardware",
"monitor.virtual_exec" => "hardware",
"vhv.enable" => "TRUE",

On some hosts, these settings cause the VM to fail to power on. In the VM's working directory, vmware.log contains the following:

2016-07-21T17:04:18.801Z| vmx| I120: Msg_Question:
2016-07-21T17:04:18.801Z| vmx| I120: [msg.cpuid.noVHVQuestion] Virtualized Intel VT-x/EPT is incompatible with this virtual machine configuration.
2016-07-21T17:04:18.801Z| vmx| I120+ Continue without virtualized Intel VT-x/EPT?
2016-07-21T17:04:18.801Z| vmx| I120: ----------------------------------------
2016-07-21T17:04:18.801Z| vmx| I120: MsgIsAnswered: Using builtin default '_No' as the answer for 'msg.cpuid.noVHVQuestion'

There is no way I have found to override the msg.cpuid.noVHVQuestion parameter in the .vmx file. If you comment out msg.autoAnswer in the .vmx file or set it to false, the following appears in the vSphere Client:

It may be possible to detect whether or not the host will experience this problem by querying the HV Replay Capable parameter from esxcfg-info:

[root@bn17-21:~] esxcfg-info | grep '\-HV Replay Capable'
         |----HV Replay Capable.....................................false

On hosts which don't experience this problem, the value is true.