nova live-migration starts instance on destination hypervisor but then fails with ERROR in Red Hat OpenStack Platform 10

Issue

nova live-migration starts instance on destination hypervisor but then fails with ERROR in Red Hat OpenStack Platform 10

nova show of the instance shows:

| fault                                | {"message": "Conflict updating instance aaaaaaaa-aaaa-aaaa-aaaa-aaaaaaaaaaaaa. Expected: {'task_state': [u'migrating']}. Actual: {'task_state': None}", "code": 500, "details": "  File \"/usr/lib/python2.7/site-packages/nova/compute/manager.py\", line 204, in decorated_function |
|                                      |     return function(self, context, *args, **kwargs)       

The instance is running under virsh list on the destination compute node but under nova show the instance can be seen as in ERROR state. The instance hypervisor is reported as the source hypervisor under OS-EXT-SRV-ATTR:hypervisor_hostname.

After the failed migration, the instance can be reset with nova reset-state (...). However, this does not stop the instance on the destination hypervisor and executing nova start will now cause a duplicate instance to be spawned on the source hypervisor.

See Diagnostic Steps for further details.