Instead of having long-lived OpenShift clusters, some users create and destroy OCP clusters very often, very rapidly. Even if they configure the CMMO the send data every 1 hour, there's there's the risk that up to 59 min of Prometheus metrics are lost, thus potentially creating data gaps.

Ideally, we should have some "flush" mechanism that users could trigger to force the operator to gather the last set of metrics and push them to RH (or write them locally in the case of restricted network users). Once this is done and users get confirmation the data has been sent/stored, users could safely destroy the cluster.