Hive mapping of compute and infra nodes to worker nodes about hive

[dgoodwin]

Concept no longer formally exists in OpenShift 4.0, and many are outright opposed to it. You would have to roll your own solution, and that would be blocked today because you cannot have more than one worker machine pool yet, and you cannot apply specific labels/taints to machine pools. (both of which would be reasonable enhancements to request)

CC @twiest

[smarterclayton]

Worker needs to be renamed compute before we ship 4.0. At this point, infra should be a supported configuration but we are considering not having it as the default.

[smarterclayton]

@aaronlevy @crawford ^ did you also expect this or have a different expectation?

[dgoodwin]

Worker is getting renamed to compute and master to control plane in openshift/installer#1033. We will do the same in Hive once it's ready.

Hive will eventually support adding additional machine pools and tagging them. See https://jira.coreos.com/browse/CO-156