We should split out the network-diagnostics controller from the CNO and incorporate it in the network-tools, as such we would be able to have a daemonset which launches a dedicated set of scripts existing in the network-tools image and then proceeds to launching the network-diagnostics controller. This would allow us to automatically log which network dependent rpms are on the node - during upgrades we would be able to compare version N to N+1 and verify that the rpm versions match what we expect.  

This would also mean that in the CNO we re-write the deployment spec currently existing for the network-diagnostics component to specify the scripts we want to launch when the components starts.