Description of problem:
Killing the qemu process on the destination host while the VM is migrating causes the VM to get stuck in Migration Source* in VDSM although in the source host the VM is 'running' in libvirt.

Version-Release number of selected component (if applicable):


How reproducible:
VDSM: 4.17.5
Libvirt: 1.2.18 (1.fc22)
QEMU: 2.4.0 (2.fc22)

Steps to Reproduce:
1. Migrate a VM
2. Kill the qemu process on the destination host
3.

Actual results:
VM stays in MigratingFrom status in the engine.

vdsClient 0 list table:
9397bed3-70a4-48bc-a0e5-77ff5a212557   1812  mmm  Migration Source* 

virsh > list
 Id    Name                           State
----------------------------------------------------
 2     mmm                            running

Expected results:
The VM should return to status UP on the source.

Additional info:
The following exception is repeating in VDSM log:
virt.vm::(_getRunningVmStats) vmId=`9397bed3-70a4-48bc-a0e5-77ff5a212557`::Error fetching vm stats
Traceback (most recent call last):
  File "/usr/share/vdsm/virt/vm.py", line 1384, in _getRunningVmStats
    vm_sample.interval)
  File "/usr/share/vdsm/virt/vmstats.py", line 43, in produce
    disks(vm, stats, first_sample, last_sample, interval)
  File "/usr/share/vdsm/virt/vmstats.py", line 262, in disks
    interval))
  File "/usr/share/vdsm/virt/vmstats.py", line 291, in _disk_rate
    (last_sample['block.%d.rd.bytes' % last_index] -
KeyError: 'block.1.rd.bytes'