Description of problem: For now, we still use the upstream images of Etcd-operator/Prometheus-operator in the OLM. But, we should replace them with their downstream images. One of hardcode items in installation: https://github.com/openshift/openshift-ansible/blob/master/roles/olm/files/rh-operators.configmap.yaml#L5819 Version-Release number of selected component (if applicable): the latest openshift-ansible master branch How reproducible: always Steps to Reproduce: 1. Install the OLM via the openshift-ansible. 2. Create an operator via the OLM, for example, Etcd-operator 3. Check the image of the Etcd-Operator Actual results: The image of Etcd-operator from upstream, for example: quay.io/coreos/etcd-operator@sha256:c0301e4686c3ed4206e370b42de5a3bd2229b9fb4906cf85f3f30650424abec2 Expected results: We should config the downstream images for the operators in the OLM. Additional info: As sync this issue with the developers & PM in the email, we won't use the downstream image of the operators in the OLM for 3.11. So, I file the target release to 4.0, correct me if I'm wrong.
This is being taken care of by the marketplace pipeline. I don't think this is a bug for OLM.
Evan, > This is being taken care of by the marketplace pipeline. I don't think this is a bug for OLM. Sure, I agree. I think the operator owners are responsible for this. So I transfer it to the Etcd-Operator team. @Junqi Please feel free to open another bug to trace the Prometheus-operator downstream issue if you needed.
cluster monitor has prometheus-operator downstream images, as the prometheus operator configured from web console, from Comment 1, it is also managed by the marketplace pipeline, no need to file bug
Junqi, > as the prometheus operator configured from web console, from Comment 1, it is also managed by the marketplace pipeline, no need to file bug I'd like to clarify this. Yeah, the Prometheus operator is managed by the marketplace pipeline. But, the operator owners are responsible for managing their image source. So, if your operator image still not based on the RHEL, I guess you'd better open a but to trace it in your component or ART team.