Commit 8fb12737 by Kevin Falcone Committed by GitHub

Merge pull request #3384 from edx/jibsheet/active-instances-inventory

The dynamic inventory script used for refresh-course-metadata
parents e50215e5 919ab4b9
#!/usr/bin/env python
"""
Build an ansible inventory list suitable for use by -i by finding the active
Auto Scaling Group in an Elastic Load Balancer.
If multiple ASGs are active in the ELB, no inventory is returned.
Assuming a single active ASG is found, a single machine is returned. This inventory
is generally used to target a single machine in a cluster to run a cmomand.
Typical reponse:
10.2.42.79,
Typical use
ansible -i $(active_instances_in_asg.py --asg stage-edx-edxapp) -m shell -a 'management command'
"""
import argparse
import botocore.session
import botocore.exceptions
from collections import defaultdict
from os import environ
class ActiveInventory():
profile = None
def __init__(self, profile, region):
self.profile = profile
self.region = region
def run(self,asg_name):
session = botocore.session.Session(profile=self.profile)
asg = session.create_client('autoscaling',self.region)
ec2 = session.create_client('ec2',self.region)
groups = asg.describe_auto_scaling_groups()
matching_groups = [g for g in groups['AutoScalingGroups'] for t in g['Tags'] if t['Key'] == 'Name' and t['Value'] == asg_name]
groups_to_instances = {group['AutoScalingGroupName']: [instance['InstanceId'] for instance in group['Instances']] for group in matching_groups}
instances_to_groups = {instance['InstanceId']: group['AutoScalingGroupName'] for group in matching_groups for instance in group['Instances'] }
# We only need to check for ASGs in an ELB if we have more than 1.
# If a cluster is running with an ASG out of the ELB, then there are larger problems.
active_groups = defaultdict(dict)
if len(matching_groups) > 1:
elb = session.create_client('elb',self.region)
for group in matching_groups:
for load_balancer_name in group['LoadBalancerNames']:
instances = elb.describe_instance_health(LoadBalancerName=load_balancer_name)
active_instances = [instance['InstanceId'] for instance in instances['InstanceStates'] if instance['State'] == 'InService']
for instance_id in active_instances:
active_groups[instances_to_groups[instance_id]] = 1
if len(active_groups) > 1:
# When we have more than a single active ASG, we need to bail out as we don't know what ASG to pick an instance from
return
else:
active_groups = { g['AutoScalingGroupName']: 1 for g in matching_groups }
for group in active_groups.keys():
for group_instance in groups_to_instances[group]:
instance = ec2.describe_instances(InstanceIds=[group_instance])['Reservations'][0]['Instances'][0]
if 'PrivateIpAddress' in instance:
print "{},".format(instance['PrivateIpAddress'])
return # We only want a single IP
if __name__=="__main__":
parser = argparse.ArgumentParser()
parser.add_argument('-p', '--profile', help='The aws profile to use when connecting.')
parser.add_argument('-l', '--list', help='Ansible passes this, we ignore it.', action='store_true', default=True)
parser.add_argument('--asg',help='Name of the ASG we want active instances from.', required=True)
args = parser.parse_args()
region = environ.get('AWS_REGION','us-east-1')
ActiveInventory(args.profile,region).run(args.asg)
Markdown is supported
0% or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment