Deployment has failed after adding 'cinder' role to 'controller' node

[Sergii Rizvan]

Detailed bug description:
 After adding 'cinder' role to 'controller' node on already deployed environment redeployment is failed with:
Error
Deployment has failed. All nodes are finished. Failed tasks: Task[restart-haproxy/1] Stopping the deployment process!

Here is the part of puppet-apply.log: http://paste.openstack.org/show/601819/

Steps to reproduce:
 1. Deploy environment with 3 controllers and 1 compute node.
 2. Log in to Fuel IU. Go to environment -> Nodes.
 3. Choose some of controller nodes and click 'Edit roles'.
 4. Add 'Cinder' role to the node.
 5. Go to Dashboard tab and click "Deploy Changes"
 6. Deployment has failed with Failed tasks: Task[restart-haproxy/1]

Reproducibility:
 Always.

Description of the environment:
 Fuel 11
 ISO: http://seed.fuel-infra.org/fuelweb-iso/fuel-11.0-community-505-2017-02-24_05-05-00.iso.torrent?from=landing
 Env: Controllers: 3 nodes, Computes: 1 node.

Additional information:
 I tried to do the same thing with next environment: 1 controller, 1 compute; and redeployment was successful.

[Oleksiy Molchanov (omolchanov)]

Please attach diagnostic snapshot.

[Sergii Rizvan (srizvan)]

Diagnostic snapshot: https://drive.google.com/a/mirantis.com/file/d/0B6XhgPR4hQqCVVo1S21zVDNKelU/view?usp=sharing

[Oleksiy Molchanov (omolchanov)]

We already have a patch - https://review.openstack.org/#/c/436514/