Cinder volume can not be used under FC backend and single path environment.

Issue

• If volume detach operation, e.g. nova volume-detach command is executed or a Cinder volume attached instance is deleted, is executed, the detach operation is completed but the device for cinder volume connection on Compute node is not deleted. Thus, next volume attach operation causes wrong device size detection then I/O error occurs in an instance because the old device is not initialized properly. And other symptom is that the old device information remains while volume attach and detach operations are repeated. This issue can't be avoidable. So Cinder volume which is FC backend and single path is not usable.
• It should be noted that Red Hat OpenStack Platform 8 or later does not have this issue because they have a different implementation of volume detach operation.