Comment 5 for bug 1254772

Looked this over with sdague, we're thinking the problem is how test_server_rescue is doing two volume creates in the setUpClass method and then when it tears down, it doesn't wait for the volumes to be deleted, it just moves on. When you have tests running in parallel, this could cause cinder to backup and then we get racy failures on volume creation.

Also, out of the 7 test cases, only two of them use the volumes, and really those are isolated, so you only need to create one volume for the entire test class and that's only in two of the seven test cases, and then we can do that in the test cases themselves and cleanup with an addCleanup call inline.