CX4-240 Undead hosts eating hba wwpns and new hosts definitions

I'm using a CX4 with a Cisco UCS chassis.  I'm trying to reassign port worldwide names for each blade which have been previously assigned to other EMC hosts in the chassis to new hosts and assign them to individual storage groups.  Unfortunately the previous host definition won't stay dead.  I go into the connectivity status wizard, edit the wwns to assign them to new hosts, then connect the hosts and LUNs to a storage group.  Everything goes swimmingly for about five minutes until the undead host returns from the grave, takes over the worldwide names, destroys the new host definition and inserts itself into the storage group. 

I've tried editing the entries in the connectivity status wizard, I've unzoned the entire UCS from the EMC and tried to deregister the undead host as follows

naviseccli -address 1.2.3.4.port -removehba -host dracula

I've tried using the deregister button in the wizard on all of the WWNs attached to the undead monster while it was un-zoned.  At random, it keeps coming back and taking over worldwide names.  I've rebooted both management servers.  The only thing I've not tried thus far is rebooting the entire array, on SP at a time.

Any suggestions that don't include fire axes, flamethrowers, or reliable storage subsystems from other vendors (which are all already on the table) would be greatly appreciated!  Please and thank you for any help.