I am trying to use Route53 for automatic failover. The problem I am having is that even though the CloudWatch alarm I am using in the health check is in an Alarm state, the health check is still reporting as healthy, and the DNS is still resolving to the faulty host (I'm simulating this by stopping the instance - I've set CloudWatch to treat missing metrics as an alarm). See screenshots for the configuration of my alarm.
As you can see, this health check is reporting as healthy: