In order to deploy your Job in cluster mode, you need to have grouped your physical servers in a virtual server via the Virtual Servers page, as explained in Configuring virtual servers.
if one of the physical servers of a specific virtual server has the best rate, this physical server will be used for the executions of the tasks.
if all physical servers have the same rate, the first physical server of the virtual server will be used for the executions of the tasks.
For more information on the server rates (based on disk space, CPU and RAM usage, etc.) which determine which physical server to be used for the executions and how to edit these weight values, see Configuring execution servers and Configuring the indicators which determine which server to be used for load balancing.
This clustered server will be used to distribute the incoming requests (Job generation, deployment and execution) between the physical servers and to ensure that these requests are processed even if one physical server is down.