Comment 0 for bug 1638683

Rebooting infra nodes is a time consuming task, more than it should be.
I like to change the boot procedure in the following way :

- General delay of 1s
- Leave the rabbitmq and galera container in the onboot group
- Move all other containers out of all lxc groups so they get treaded as NULL group and started past galera/rabbit

That way the node is operational 7 minutes faster