Once we have multiple clustered REST servers in place, there needs to be a checklist of common failure situations that can occur, and the UI needs to be tested against the REST servers in a failed state.