The "0000_90_olm_00-service-monitor.yaml" manifest containing RBAC for Prometheus to scrape OLM namespace is omitted https://github.com/openshift/hypershift/blob/e9594f47570b557877009b607d26b9cb4a34f233/control-plane-operator/controllers/hostedcontrolplane/cvo/reconcile.go#L66

But "0000_50_olm_06-psm-operator.servicemonitor.yaml" containing a new OLM ServiceMonitor that was added in https://github.com/openshift/operator-framework-olm/pull/551/files is still deployed, which make Prometheus logs failures: see https://issues.redhat.com/browse/OCPBUGS-36299

    

    Check Prometheus logs in any 4.17 hosted cluster
    

    1.
    2.
    3.
    

    Prometheus shouldn't be asked to discover/scrape targets without giving it the appropriate RBAC
   
   If that new ServiceMonitor is needed, the appropriate RBAC should be deployed, if not, the ServiceMonitor should be omitted.

    Maybe Hypershift should use an opt-in approach instead of opt-out for OLM resources, to avoid such issues in the future.