Detecting Missed or Abandoned Migrations when Plays Fail

The current playbooks will not recover happily if a database migration is queued, the job fails, and then is run again.

Need to figure out how to detect this state and react appropriately.