I wan't to kill a specific instance when some of it's hardware consumption metrics reaches a certain level. If i create an alarm for the scaling group (setting maximum cpu consumption threshhold to >=50 for example) it will kill the oldest instance - not the one that is misbehaving. One way to kill a specific instance is to create an alarm for the instance instead of creating it for the scaling group. However, if i launch a new instance (with autoscaling), this new instance will not have the alarm.
Is there a way to launch instances with preconfigured ec2 level alarms?