Comment 33 for bug 1696125

It doesn't look like a timeout issue. When the test is going to fail, you see these messages in the logs:

2017-09-19 14:13:01.172 DEBUG oslo.service.loopingcall [-] Invoking nova.virt.libvirt.guest._do_wait_and_retry_detach; retry count is 6. from (pid=611) _func /usr/local/lib/python2.7/dist-packages/oslo_service/loopingcall.py:393
...
2017-09-19 14:13:23.765 DEBUG oslo.service.loopingcall [-] Invoking nova.virt.libvirt.guest._do_wait_and_retry_detach; retry count is 7. from (pid=611) _func /usr/local/lib/python2.7/dist-packages/oslo_service/loopingcall.py:393

When the test passes, you don't see these messages, so if retries start happening, the detach will eventually fail.