At 3.30pm today we were alerted to multiple unhealthy nodes in our cluster. New hosts were provisioned and traffic was rerouted to healthy nodes. As part of this process some downtime will have been experienced by a small percentage of our traffic.