It should be possible to create a clustered singleton by just adding the @Clustered annotation.

@Singleton
@Clustered
public class MyScheduler {
}

Care should be taken that the life-cycle callbacks are only invoked once on one node, the other nodes will only receive the state of the singleton after post-construct.

Concurrency management of the singleton should happen cross-cluster, so that a singleton with read lock can be concurrently invoked on any node. After existing a method with a write lock the state must be synchronized across the cluster before new invocations are processed.