We've been using a db-follows-branch system with Fabric. Devs can run "fab update_db" to get a new dump of production-based testing data, and create a database named after the current branch. Still a bit manual, but it really helps in situations like that.