This is a clone of issue OCPBUGS-8044. The following is the description of the original issue:
rhbz#2089199, backported to 4.11.5, shifted the etcd Grafana dashboard from the monitoring operator to the etcd operator. During the shift, the ConfigMap was renamed from grafana-dashboard-etcd to etcd-dashboard. However, we did not include logic for garbage-collecting the obsolete dasboard, so clusters that update from 4.11.1 and similar into 4.11.>=5 or 4.12+ currently end up with both the obsolete and new ConfigMaps. We should grow code to remove the obsolete ConfigMap.
4.11.>=5 and 4.12+ are currently exposed.
1. Install 4.11.1.
2. Update to a release that defines the etcd-dashboard ConfigMap.
3. Check for etcd dashboards with oc -n openshift-config-managed get configmaps | grep etcd.
Both etcd-dashboard and grafana-dashboard-etcd exist:
Another example is 4.11.1 to 4.11.5 CI:
Only etcd-dashboard still exists.
A new manifest for the outgoing ConfigMap that sets the release.openshift.io/delete: "true" annotation would ask the cluster-version operator to reap the obsolete ConfigMap.