In a resilient deployment with multiple Meeting Servers, we recommend deploying a database cluster of more than 2 nodes. (There is currently a limit of 5 database nodes in a cluster.)
When you create a cluster of Meeting Server databases, the first node added to the cluster will be elected as the Master database. The additional nodes come on as slaves. If you only have two nodes in the cluster and there is a failure on the master node, the remaining slave doesn't have any way to check to see if it's safe to promote itself to being the master and continue servicing requests. By adding a third database node, the database cluster now has a way to help determine where the failure is, and if it's safe to elect a new master, so the Meeting Server operations can continue uninterrupted.