Retention procedure prunes the number of cluster backups created according to the criteria specified within the CR ( see https://github.com/openshift/enhancements/pull/1370 )

Retention is being managed by Cluster-etcd-operator https://github.com/openshift/cluster-etcd-operator/blob/master/pkg/cmd/prune-backups/prune.go 

Using hostPath as storage, the backups are being created on one of the master nodes local volume. 

Although the intention is keep the backups on only one master node, it was found that the backups are being created on any master node where the backup pod is being scheduled. Moreover, the retention keeps running on the initial master node where the first backup has been created, and not on any other master nodes. Consequently, the backups are not being pruned on all the master nodes.