Description
Pods can be given topology spread constraints, and this does what our stripe scheduler does for us. A spread constraint can work across clusters too as the node selector keys can match, whereas with explicit scheduling the values will always differ (this is especially pertinent for tech like velero and portworx replication).
The big question, is whether we can tell where we were scheduled in order to populate Couchbase server groups.
Requires Kubernetes 1.19+, so this isn't relevant for a few releases yet.