When you cannot access the controller, usually that means that TCP 48000 is blocked from where you are running the Infrastructure Manager (unless you are using the Admin Console) or from the hub if a tunnel is involved. Sometimes with a cluster that can be caused by the wrong IP being advertised to the hub by the robot. If you click on the hub in Infrastructure Manager, you should see the IP in the robot list. Make sure it is right; if not, you have to edit it in robot.cfg.
Normally you would install the robot on the second machine as well. I am not sure if you can get Nimsoft to be part of the cluster rather than part of each node. That may be possible, but I have never tried it.