cinder volume create with bad availability_zone should fail immediately

[stephen mulcahy]

the cinder api accepts requests for volume creation specifying a non-existent availability_zone with the volume created going into the error state once the backend recognises that the availability zone doesn't exist. This error should be caught at the api and an error immediately returned to the user.

To reproduce in devstack:

cinder create --availability_zone badaz 1

(request accepted, volume creation begins)

cinder show <volume uuid>

volume in error state

[Xiaoxi Chen (xiaoxi-chen)]

Seems the request will failed by scheduler actually, will not go to backend.
I think this make sense ,no reason for API to keep a complete Availability-zone list

[stephen mulcahy (stephen-mulcahy)]

The user experience would be better if they received a failure on submitting the create volume request rather than a volume in an error state some time after submitting the request - with no indication of why the volume is in an error state.

[stephen mulcahy (stephen-mulcahy)]

this information should be available from the services table so the api should be able to do some sort of validation

[OpenStack Infra (hudson-openstack)]

Fix proposed to branch: master
Review: https://review.openstack.org/32422

[OpenStack Infra (hudson-openstack)]

Reviewed: https://review.openstack.org/32422
Committed: http://github.com/openstack/cinder/commit/1f994f2589411e80852650025a5a8e6cb0dc8c85
Submitter: Jenkins
Branch: master

commit 1f994f2589411e80852650025a5a8e6cb0dc8c85
Author: Haomai Wang <email address hidden>
Date: Mon Jun 10 23:20:26 2013 +0800

    Add availability zone checking in the api service

    This patch adds availability zone checking for api request period.
    If the availability zone the request specified isn't exist in the
    Cinder services, the response returned to the user will give an
    error immediately.

    Fixes bug: 1185714

    Change-Id: I842eec7b500c3ff6a7e5d8fc5e397d61617b0eea