Replace a primary controller failed: "command ... has been running for more then 10 seconds!"

[Dennis Dmitriev]

Scenario:
            1. Revert snapshot deploy_neutron_tun_ha with 3 controller
               and 2 compute nodes
            2. Add 'cinder' role to a new slave
            3. Deploy changes
            4. Remove the 'cinder' node
               Remove a 'controller' node
               Add 'controller'+'cinder' multirole to a new slave
            5. Deploy changes

Result: step 5 failed with the following errors in puppet on the original primary controller:

node-1 2015-09-10T23:55:00.449056 err: Could not prefetch keystone_service provider 'openstack': Command: 'openstack service list --quiet --format csv --long' has been running for more then 10 seconds!

node-1 2015-09-10T23:56:10.487462 err: Command: 'openstack service create --format shell identity --name keystone --description OpenStack Identity Service' has been running for more then 10 seconds!

node-1 2015-09-10T23:56:10.490786 err: (/Stage[main]/Keystone::Endpoint/Keystone::Resource::Service_identity[keystone]/Keystone_service[keystone]/ensure) change from absent to present failed: Command: 'openstack service create --format shell identity --name keystone --description OpenStack Identity Service' has been running for more then 10 seconds!
....