zmq driver broken with cinder multi-backend

[James Page]

The cinder-scheduler service resolves the host to send messages to as:

  juju-jpos-machine-5@cinder-ceph

This results in a tcp connection being opened to a non-existant host:

2014-10-16 09:51:07.086 12389 DEBUG oslo.messaging._drivers.impl_zmq [-] Connecting to tcp://juju-jpos-machine-5@cinder-ceph:9501 with PUSH __init__ /usr/lib/python2.7/dist-packages/oslo/messaging/_drivers/impl_zmq.py:145
2014-10-16 09:51:07.087 12389 DEBUG oslo.messaging._drivers.impl_zmq [-] -> Subscribed to None __init__ /usr/lib/python2.7/dist-packages/oslo/messaging/_drivers/impl_zmq.py:146
2014-10-16 09:51:07.088 12389 DEBUG oslo.messaging._drivers.impl_zmq [-] -> bind: False __init__ /usr/lib/python2.7/dist-packages/oslo/messaging/_drivers/impl_zmq.py:147

and no message ever being sent - the volume sticks in creating state.

[OpenStack Infra (hudson-openstack)]

Fix proposed to branch: master
Review: https://review.openstack.org/129185

[OpenStack Infra (hudson-openstack)]

Change abandoned by Victor Sergeyev (<email address hidden>) on branch: master
Review: https://review.openstack.org/129185
Reason: This patch doesn't seems to be maintained.

[Ainur Shakirov (ainur-shakirov-tt)]

OpenStack Ocata/Cinder 10.0.0: the bug is still exist.

[Ainur Shakirov (ainur-shakirov-tt)]

Note: When I tries to create the volume cinder-wsgi starts load CPU to 100% and the volume creates with "Error" status.