With 2 controllers, running OS 8, without an MM, you can configure a standalone with another standalone as a VRRP backup. This is similar to Master/Master in OS 6 and earlier. VRRP needs to be configured and database synchronization also needs to be configured as part of VRRP.
With just 2 controllers, you cannot deploy what was referred to in OS 6 and earlier as Master/Local and have failover.
OS 8 has 3 install designs, listed below in no particular order.
The first is Mobility Master (MM) with Mobility Controllers (MCs). The MM is strictly a management/configuration machine, which distributes the configuration out to the MCs. For redundancy, you would need two or more MCs. An MM environment supports clustering, which provides failover with stateful preservation, meaning clients essentially do not lose connection or state on an MC failure.
The second is standalone. This is a controller running by itself. Standalone is where the configuration is done, and APs and client traffic is processed by the standalone controller. If you would like redundancy, you can deploy a second standalone, with VRRP and database synchronization, and you will have an Active/Standby failover environment. Any failure would disconnect the clients which will have to reconnect. It is reasonably quick (depending upon the number of APs and clients), but stateful connections will be lost.
The third is Master Controller Mode (MCM). This is kinda like the MM mode, but instead of an MM being the management platform, a controller is the managment platform (we refer to this similar to OS 6 and earlier as the Master). In MCM the Master controller is ONLY a management machine. In OS 6 and earlier, a Master controller could terminal APs; that is not the case with OS 8. So with 2 controllers, with MCM, one is the Master and the other is the MC (similar to a "local" controller in previous versions of ArubaOS). The difference is that in previous versions, an APs and clients could connect to the local, and it the local failed, they could then connect to the Master. You can't do this with OS 8. So in order to achieve failover in an MCM environment, you would need a Master and 2 MCs. You could implement VRRP, LMS-IP/Backup LMS-IP, or Fast Failover also known as High Availability (HA). All three of these failover methods will have clients lose their stateful connections.
I hope this helps,
I hope this helps,