I have two Asterisk+FreePBX servers that share a common database. The services are active on one server at a time, and in the event of a failure of the main server, the services are then started on the failover node. Currently, they use a master-slave (replica) database architecture. This configuration works well.

Topology database.png

However, I?m considering reconfiguring this topology to a master-master setup with a floating IP address. In that case, all reads and writes would be performed on a single node database. In the event of a database server failure, the system would automatically switch to the other active one. From my perspective, this would create a relatively simple HA topology for the VoIP service.

But would implementing scenario 2 be a good idea?
Wouldn't this cause potentially any data consistency issues?
Am I correct in assuming that once the server comes back online, all the changes will automatically synchronize?

IP failover only happens in case of a failure (that's how I configured it). So even if the node that failed comes back online, the IP address will not be automatically switched back to it. This configuration ensures high stability of my entire topology.