With the new code for a endpoint to work in a clustered environment it should have a unique name. This implies we cannot use anonymous endpoints. For example we cannot use an endpoint inside a send mediator without specifying a name. From a users perspective they expect a synapse configuration working in a non-clustered environment to work in a clustered environment as well. Requiring to tweak the synapse configuration when a user migrate to a clustered environment is not that user friendly.