Error with more than one cinder-volume with qpid as rpcbackend

[renminmin]

The grizzly version openstack use qpid as rpcbackend.
cinder service deployed it on two nodes(node1 and dev202)
node1: cinder-api cinder-scheduler cinder-volume, keystone, mysql, qpid
node2: only cinder-volume

rpc_backend = cinder.openstack.common.rpc.impl_qpi

qpid-tool watch the queue status (list queue active):
    252353 07:50:56 - 248315.cinder-scheduler
    252354 07:50:56 - 248315.cinder-scheduler.node1
    252355 07:50:56 - 248315.cinder-scheduler_fanout_e0ef7071e1b744769df5f06dae595550
    252479 07:52:27 - 248315.cinder-volume
    252480 07:52:27 - 248315.cinder-volume.node1
    252481 07:52:27 - 248315.cinder-volume_fanout_df149604027d49fabd8853f3acb7e997
    252549 07:52:49 - 248315.cinder-volume.dev202
    252550 07:52:49 - 248315.cinder-volume_fanout_4bab111c0de74d8285b12ba4990d9ec9

then I stop cinder-volume service on node1
list queue active:
    252353 07:50:56 - 248315.cinder-scheduler
    252354 07:50:56 - 248315.cinder-scheduler.node1
    252355 07:50:56 - 248315.cinder-scheduler_fanout_e0ef7071e1b744769df5f06dae595550
all of cinder-volume queues were deleted.
After this operation, cinder create volume status became creating, even cinder-volume on dev202 is still active but no queue message receive.
However, cinder-volume on dev202 still report its status to cinder-scheduler.

More discussion link:
https://answers.launchpad.net/cinder/+question/229456