During a planned upgrade of a specific OpenShift node, a substantial part of the AMQ Online infrastructure is lost. This happens because the AMQ Online operator does not appear to provide a way to configure the properties that would allow OpenShift to redistribute services in advance of the upgrade. As a minimum, it ought to be possible to specify a "maxUnavailable" level on broker pods. Better, though, would be the ability to configure pod affinity rules, that would distribute brokers such that the services don't end clustered on the same node.