Hi,
I have seen similar problems during my first installation of 3.1.0.7
What I have defined is now:
in each resource group a different virtual ip-address
in each resource groups own dsm.opt:
tcpclientaddress <the virtual one>
tcpclientport <a different number in each occurance>
backupregistry no     /* registry will only be backed up from the cluster nodes
*/

then i logged in once manually with each option file e.g.
>dsmc -optfile=F:\adsm\dsm.opt -password=mypasswd

the schedulers starting mode should be manual - it will be started from the
resource-manager when the resource is set online on the cluster-node
Hope this helps
Regards, Karl