Background:

Right now it is really difficult to figure out which node is the "problem node" in certain states, because all the MachineConfigPool shows you is the nodes and their counts.

You can know there is a node with a problem, but it's hard to know which node is the problem without inspecting all the node conditions (and if it's something that doesn't surface on `oc get node` – like DiskPressure – woe be to you)

There are customers that have a ton of nodes, so we have to be mindful of etcd space and making our pool output noisy.

I think it makes sense to conservatively only show nodes with "problems", but we need to be mindful that if we consider "not updated yet" to be a problem, the list will reset every time we roll out a new machineconfig.

Some Options:

• Add node states to the pool CRD
• New CRD containing pool states
• Add some sugar to `oc` to make state easier to retrieve

Goal:

• Figure out an efficient/non-noisy/non-annoying way to clearly show which nodes are in bad states

Done When:

• We can write the story to do the actual work of adding the node state display