The following debug message can be seen when doing jmxmp based cascading with JDMK 5.1 FCS (refer to 6328682 for Java code for reproducing as well as full output) :

FINER: Failed to register a listener to the mbean server: the client will not do clean when an MBean is unregistered
java.lang.ClassCastException: [Ljava.lang.Integer; cannot be cast to java.lang.Integer
at javax.management.remote.generic.ClientIntermediary$GenericClientNotifForwarder.addListenerForMBeanRemovedNotif(ClientIntermediary.java:902)
at com.sun.jmx.remote.opt.internal.ClientNotifForwarder.init(ClientNotifForwarder.java:626)
at com.sun.jmx.remote.opt.internal.ClientNotifForwarder.addNotificationListener(ClientNotifForwarder.java:94)
at javax.management.remote.generic.ClientIntermediary.addNotificationListener(ClientIntermediary.java:586)
at javax.management.remote.generic.GenericConnector$RemoteMBeanServerConnection.addNotificationListener(GenericConnector.java:540)
at com.sun.jdmk.remote.cascading.proxy.ProxyCascadingAgent.start(ProxyCascadingAgent.java:394)
at com.sun.jdmk.remote.cascading.CascadingService$MountPoint.mount(CascadingService.java:134)
at com.sun.jdmk.remote.cascading.CascadingService.mount(CascadingService.java:260)
at Defect.run(Defect.java:48)
at Defect.main(Defect.java:20)

The semantic it has is frightening as such message is theorically only informative : "no clean" means potential resources leak.
Is there really such leakage risk ?
Is it possible to suppress the root cause of that ClassCastException ?
If we have no choice but seeing an exception there, please clarify what it means.
No ClassCastException occurs when running the same code with rmi.