[MOS]cinder-volume autostart fail

[Rodion Promyshlennikov]

Deployment with Vbox help scripts without any cinder-volume daemons started.
Configuration:
   4 slaves from launch8Gb script with extended memory (1,5GB incremented up to 3GB) for controllers, config.sh attached.
steps to reproduce:
 1. MOS 7.0
 2. use vbox helper scripts
 3. start deployment with launch_8GB.sh (see attached edited config.sh)
 4. select basic config(cinder vs ceph, basic networking, etc by default options) for deploy
 5. select 2 nodes with 3GB as controllers
 6. select other 2 nodes as compute nodes
 7. use 3rd "free" net interface to work with cinder (volumes) on all nodes
 8. cinder-volume should be on all 4 nodes
 9. start deploy

expected result:
1. deployed cloud without any issues.

observed results:
1. cinder-volume services stopped on controllers and compute-nodes.
after manual start this services, functional and other tests have passed, cloud was working.

workaround:
start manually on all nodes.

[Rodion Promyshlennikov (rpromyshlennikov)]

Deployments steps.

[Rodion Promyshlennikov (rpromyshlennikov)]

You can download master node snapshot from here, ~250MB:
https://drive.google.com/file/d/0B-QiiEr4w70UR2EwNFJnQkFidWc/view?usp=sharing