In a Galera-based system, failover is handled automatically. Since it is a multi-master topology, any node can handle read and write operations. If one node fails, the remaining nodes simply continue to serve traffic without a service disruption.
The best part is the recovery process. Once the failed node comes back online, it automatically rejoins the cluster and syncs up with the latest state from the other nodes. For mission-critical apps where you cannot afford downtime, a managed cloud platform like AccuWeb.Cloud is a solid choice because they include auto-healing and live replication features as part of the managed service.