Problem

HyperShift introduces a new flavor of OpenShift where the control-plane and data plane are loosely coupled. This architectural change impacts many of the standard standalone OCP components, one of these affected components is monitoring.

Customers are interested in making sure the control-plane availability meets certain SLAs. The availability of an OpenShift cluster is usually tied to important service endpoints such as the API-server. Without a monitoring stack, it is hard to define those SLAs.

Proposal

 There are multiple ways to go about monitoring the control-plane in HyperShift. One way to do so is to capitalize on HyperShift's design, where the control-plane is running as pods and can be treated as user-workloads in the management cluster (the cluster where all the control planes for all clusters are hosted). 

That said, re-use the existing in-cluster monitoring stack, one approach could be to define service monitors for SLA accounted services.