In the server group, there is a fail-through option. If this is off, the first server will be always queried. if the first server does not respond, it will be taken out of service and the second server will be used.
If fail-through is ON, the first server will be queried. If we get a reject from the first server, the second server will be queried. That is the option you need.