We've had a sneaking suspicion there is a bug because this isn't the only syslog data that's unreliable