This incident has been resolved.

We had a failure in one of our availability zones that caused a database failover. The failover is complete and we are monitoring the results.