boot an instance via a remote volume created by a specified image ,when the spwan process failed for some reasons , the reschedule process disabled to recreate the instance.
to simulate the spwan process failed, an exception was raised just before the function self.driver.spwan() called by _build_and_run_instance function() in compute/manager.py file.
then excute "nova boot --flavor 1 --block-device id=my_image_id,source=image,dest=volume,bus=virtio,shutdown=removed,device=/dev/vda,bootindex=0,size=1 --nic net-id=my_net_id",created instance failed.
cinder list showed that the state of the remote volume is in-use.
It seemed that the reschedule process failed cause the volume wasn't freed.