From a technology and a setup point of view both approaches are identical. They differ in the usage of the individual nodes and the loadbalancer setup to achieve this usage. In active/active your loadbalancer distributes requests to both nodes, in active/passive to only one (normally the master node).
In my experience it should be sufficient for 90% of all usecases to use active/passive, as it will give you the same amount of high-availability as the active/active setup.
Jörg