get_all_groups returns invalid objects when any ASG is attached to a target group

[tangentspace]

We recently started using Application Load Balancers, and many legacy Python scripts that still use boto 2.x started failing. Any ASG that's attached to a target group results in a malformed object coming back from the get_all_groups function, here is the __dict__ attribute from one such object (note the many missing values such as min_size, name, and desired_capacity):

{'autoscaling_group_arn': u'arn:aws:autoscaling:us-west-2:012345678910:autoScalingGroup:b72815ad-fafa-42e5-9f66-9d6bd4608a1c:autoScalingGroupName/devops-develop-jenkins-ng',
'availability_zones': [u'us-west-2a', u'us-west-2b', u'us-west-2c'],
'connection': AutoScaleConnection:autoscaling.us-west-2.amazonaws.com,
'created_time': None,
'default_cooldown': 300,
'desired_capacity': None,
'enabled_metrics': [],
'health_check_period': None,
'health_check_type': u'EC2',
'instance_id': None,
'instances': [Instance<id:i-af752bb7, state:InService, health:Healthy>],
'launch_config_name': None,
'load_balancers': [],
'max_size': 2,
'member': u'arn:aws:elasticloadbalancing:us-west-2:012345678910:targetgroup/devops-develop-jenkins-ng/74eda25cc67b618d',
'min_size': None,
'name': None,
'placement_group': None,
'tags': None,
'termination_policies': [u'Default'],
'vpc_zone_identifier': None}

It looks like boto has not been updated to handle a non-empty TargetGroupARNs element in the XML response (

boto/boto/ec2/autoscale/group.py

Lines 212 to 265 in 941346c

	def startElement(self, name, attrs, connection):
	if name == 'Instances':
	self.instances = ResultSet([('member', Instance)])
	return self.instances
	elif name == 'LoadBalancerNames':
	return self.load_balancers
	elif name == 'AvailabilityZones':
	return self.availability_zones
	elif name == 'EnabledMetrics':
	self.enabled_metrics = ResultSet([('member', EnabledMetric)])
	return self.enabled_metrics
	elif name == 'SuspendedProcesses':
	self.suspended_processes = ResultSet([('member', SuspendedProcess)])
	return self.suspended_processes
	elif name == 'Tags':
	self.tags = ResultSet([('member', Tag)])
	return self.tags
	elif name == 'TerminationPolicies':
	return self.termination_policies
	else:
	return
	def endElement(self, name, value, connection):
	if name == 'MinSize':
	self.min_size = int(value)
	elif name == 'AutoScalingGroupARN':
	self.autoscaling_group_arn = value
	elif name == 'CreatedTime':
	self.created_time = value
	elif name == 'DefaultCooldown':
	self.default_cooldown = int(value)
	elif name == 'LaunchConfigurationName':
	self.launch_config_name = value
	elif name == 'DesiredCapacity':
	self.desired_capacity = int(value)
	elif name == 'MaxSize':
	self.max_size = int(value)
	elif name == 'AutoScalingGroupName':
	self.name = value
	elif name == 'PlacementGroup':
	self.placement_group = value
	elif name == 'HealthCheckGracePeriod':
	try:
	self.health_check_period = int(value)
	except ValueError:
	self.health_check_period = None
	elif name == 'HealthCheckType':
	self.health_check_type = value
	elif name == 'VPCZoneIdentifier':
	self.vpc_zone_identifier = value
	elif name == 'InstanceId':
	self.instance_id = value
	else:
	setattr(self, name, value)

)

[RomekRJM]

I've also encountered that problem, I'm about to post the merge request with the fix.

[RomekRJM]

Done: #3616 . I've also added some tests, hopefuly it's going to be merged soon.

[shaharmor]

Any chance to get this fixed?

[RomekRJM]

@shaharmor I've sent that pull request 6w ago and no one picked it, Looking at 344 other unmerged patches, I'm fairly sure it's not going to be done soon, if ever. I honestly recommend switching to boto3, v2 does not seem to be actively maintained.

[vanga]

I tried to look at the code, but couldn't figure out why it was breaking if startElement function returns None.

I am just wondering, isn't the correct fix is to not make it error out when there is a new/unhandled property?

Just adding another else condition is going to only handle this particular property Target Groups, if tomorrow AWS adds one more property, will this fix break again?

I am just curious,

Thanks.

[RomekRJM]

I can imagine, that some people would actually want to make use of that particular field. Anyway, boto3 is the way to go. We've migrated our stuff to it and works like a breeze.

[SM616]

Faced with this as well while running Ansible ec2_elb module with latest version of boto. Any update on when the fix will be merged?

[locomundo]

I also ran into this issue. Is there a plan to fix it in boto2 ? I see this was filed quite a while ago...

[matisku]

Any progress on that?

[jorotenev]

Just encountered the same issue. This affects the boto_asg module of saltstack