You can not do this, as there would be manual changes needed to failover and failback.
The problem is that by deploying to the system you give it the authoritative copy of the configuration. So if you now deploy to the second master by changing the endpoint in the UI, it will also get an authoritative copy, so you can failover manually. But when the first master now comes back, there are two different authoritative copies which will cause trouble. So you would need to manually remove it from the first master before joining the cluster again or even better clean both, configure back to deploy on the first master and then deploy again as failback.
The other feature that is not ready for automatic failover/failback it the CA which you perhaps will also need in case of a failure.