How can I determine the "reason" behind the status of an EC2 instance that has just been registered on Amazon ELB? - amazon-elb

Question

I'm working on a deployment script (more specifically, an Ansible module) that registers an EC2 instance with an Amazon ELB. The script uses the Boto library.
Here's a look at the relevant part of the script:
def register(self, wait):
"""Register the instance for all ELBs and wait for the ELB
to report the instance in-service"""
for lb in self.lbs:
lb.register_instances([self.instance_id])
if wait:
self._await_elb_instance_state(lb, 'InService')
def _await_elb_instance_state(self, lb, awaited_state):
"""Wait for an ELB to change state
lb: load balancer
awaited_state : state to poll for (string)"""
while True:
state = lb.get_instance_health([self.instance_id])[0].state
if state == awaited_state:
break
else:
time.sleep(1)
(BTW the code above is from Ansible's ec2_elb module.)
So, when the instance is first registered it is 'OutOfService'. The script here "waits" for the instance to reach the state 'InService' after it has passed health checks etc.
So here's the problem: The process above is overly simplistic (which is why I'm trying to customize the module for my own purposes). The main problem I've hit is that if the load balancer is not configured to service the availability zone that the instance resides in, then the instance will remain out of service. Essentially the script above will just hang.
What I'd like to do (and that's why I'm customizing this built in module) is find a way to determine if the ELB is just waiting around for the instance to pass the healthcheck OR if there is some other reason (like an unregistered availability zone) that's causing it to remain Out of Service.
The Boto library (via the Amazon ELB API) does provide slightly more detail than state: it has a "reason" attribute which is described in the Boto docs (and also the Amazon ELB API docs) as follows:
reason_code (str) – Provides information about the cause of an
OutOfService instance. Specifically, it indicates whether the cause is
Elastic Load Balancing or the instance behind the LoadBalancer.
There is a paucity of documentation on the reason_code attribute that I could find out there, so I'm not sure a) what I can expect the possible return value to even be here, and b) what they actually mean in relation to my question above.
I think what I want to do is doable given that Amazon is able to display a detailed reason for why an instance is out of service is the management console -- and from what I understand they are dogfooding their API there.
So how/where can I find the more detailed reason behind the instance's status?

Answer 1

Ah, it's the description field of InstanceState:
description (str) – A description of the instance.
I guess that was so vague that my brain ignored it.
It also looks like the possibilities of state are two string values:
'ELB'
'Instance'
That's just from playing around with the API; it's not definitive or anything.