kvm: ssh commands during resize instance gets stuck up

Scenario:
I have an OpenStack set up (using master branch) with 2 Compute nodes, following are the Nova configurations details:
    allow_resize_to_same_host=True
    compute_scheduler_driver=nova.scheduler.filter_scheduler.FilterScheduler
    connection_type=libvirt
    libvirt_type=kvm

Note: There is no password less ssh connection setup between the two Compute nodes.

To reproduce this issue, spawn an instance, once it comes up, resize it. I used the nova CLI command to resize (nova resize d24f03d8-5757-453d-a025-1e7b36c6ef8f 2).

Branch: master

Expected Behavior: The instance should get resized.

Actual Behavior: The instance gets stuck in the following state: vm_state=resizing, task_state=resize_prep, power_state=0

On further analysis, it seems that the source Compute host has destroyed the instance and has got stuck in an 'ssh' command (in migrate_disk_and_power_off( ) method in libvirt/connection.py). Following is the code snippet:

            if same_host:
                utils.execute('mkdir', '-p', inst_base)
            else:
                utils.execute('ssh', dest, 'mkdir', '-p', inst_base)

Effectively, the instance directory on the source Compute host has the 'instance-00000003_resize' directory.
To proceed further, I killed the blocking 'ssh' command, however, it further got stuck in removing the instance directory on the remote Compute host. Here is the related code snippet:

            try:
                if os.path.exists(inst_base_resize):
                    utils.execute('rm', '-rf', inst_base)
                    utils.execute('mv', inst_base_resize, inst_base)
                    utils.execute('ssh', dest, 'rm', '-rf', inst_base)
            except Exception:
                pass

On killing this ssh command, the instance went into following state:
    vm_state=error, task_state=resize_prep, power_state=0.