We have records that indicate another instance in the load balancer was also affected. For that reason, that instance removed from the targets and that seems to have resolved the issue completely. We will ensure that instance is completely operational before bringing it back up. Sorry for the inconvenience.