Description of problem:

If a HA enabled VM is running in a Host and the host is gracefully rebooted, HA is not triggered because the engine sees it as "shutdown from within the Guest".

1) VM running
May 24 12:30:39 gsslab-24-32 systemd: Starting Virtual Machine qemu-2-DummyVM.

2) Reboot command
May 24 12:31:41 gsslab-24-32 systemd: Stopping Virtual Machine qemu-2-DummyVM.
May 24 12:31:41 gsslab-24-32 systemd: Stopping oVirt ImageIO Daemon...

3) VDSM:

2017-05-24 12:31:42,007+1000 INFO  (libvirt/events) [virt.vm] (vmId='f5fd1ce3-7b75-44e6-a670-a536b9dcd054') Changed state to Down: User shut down from within the guest (code=7) (vm:1207)

2017-05-24 12:31:43,017+1000 INFO  (jsonrpc/3) [jsonrpc.JsonRpcServer] RPC call VM.destroy succeeded in 0.01 seconds 
(__init__:515)

4) VDSM stopping

2017-05-24 12:31:44,985+1000 INFO  (MainThread) [dispatcher] Run and protect: prepareForShutdown(options=None) (logUtils:51)

5) Engine is notified

2017-05-24 12:31:43,010+10 DEBUG [org.ovirt.vdsm.jsonrpc.client.internal.ResponseWorker] (ResponseWorker) [] Event arrived from XXXXXX containing {"notify_time":4455448400,"f5fd1ce3-7b75-44e6-a670-a536b9dcd054":{"status":"Down","timeOffset":"0","exitReason":7,"exitMessage":"User shut down from within the guest","exitCode":0}}

2017-05-24 12:31:43,012+10 INFO  [org.ovirt.engine.core.vdsbroker.monitoring.VmAnalyzer] (ForkJoinPool-1-worker-2) [] VM 'f5fd1ce3-7b75-44e6-a670-a536b9dcd054' was reported as Down on VDS 'd406a6c4-8c8e-46c8-8bb6-0d4fd17cd470'(ovirt-H2)

Since it was "shut down from within the guest", HA is not triggered and the VM is not restarted elsewhere. 

This is not desirable as an unintentional reboot escapes the HA logic.

Version-Release number of selected component (if applicable):
vdsm-4.18.21-1.el7ev
rhevm-4.0.6.3-0.1.el7ev

Also reproduced on:
vdsm-4.19.10.1-1.el7
ovirt-engine-4.1.1.8-1.el7

How reproducible:
100%, as above

Actual results:
HA VM is down after host reboot

Expected results:
HA VM always up unless explicitly shut down by the administrator.