Make the application health a cron that reports back to the management console.

Little icon when health check doesn't pass. Email alert to onboarding group when health alert is created. Add filter to only show unhealthy instances.