Complex status for OpenStack clusters

Problem description
-----------------------------

Consider an OpenStack cluster with N nodes running different services for some time. The status of that cluster is Operational, the status of all nodes is also fine.
A cloud operator wants to add a few extra compute nodes to that cluster and so they go to the Fuel web UI, assign the appropriate role to some of available nodes and press the Deploy Changes button. At this moment the status of the cluster starts actually be meaningless – it cannot be Operational, because that status does not show that nodes are being configured in that cluster but at the same time it cannot be Provisioning, because the cluster continues to serve normally. If provisioning or deployment of enough number of those compute nodes fail the status of the task will be set to error. However, the cluster still serves normally.

There of course will be cases when failure of one or more tasks or nodes means failure of the entire cluster but there are dozens of scenarios when the status of the cluster will be meaningless in its current form. Specific tasks can be used for telling the status of a cluster neither.

User stories
----------------------

1. As a cloud operator one wants to see the detailed and meaningful information about the status of a cluster without having to look over a list of nodes.

2. As a developer of a Fuel components one requires a unified and centralized way of getting details of the status of the cluster without having to create a lot of extra requests and to avoid using sophisticated and different algorithms in different places and use those details to set different configurations to those components.