AWS Auto Scaling Groups: scheduled start/stop

toggleAutoScalingGroup.md

Tutorial: scheduled start/stop of EC2 instances managed by Auto Scaling Groups

If your EC2 instances in AWS are managed through Auto Scaling Groups, it is easy to schedule startup and shutdown of those instances, e.g. to save money.

This tutorial walks you through setting up an AWS Lambda function that is triggered by CloudWatch Events and automatically changes the min, max and desired instances in your Auto Scaling Group(s).

The idea is to toggle between 0 (stop) and a specifed min, max and desired amount of instances (start), so you only need a single Lambda function. The premise is that you do not touch these Auto Scaling Group settings manually, or you might make your EC2 instances nocturnal.

Create new Lambda function and Start Event

1. Open AWS Lambda in the console and click Create a Lambda function.
2. On the Select blueprint page, select Blank function.
3. On the Configure Triggers page, click the grey square and select CloudWatch Events.
4. From the Rule dropdown, select Create a new rule.
5. Enter a Rule name and Rule description.
6. For Rule type, select Schedule expression.
7. For Schedule expression, enter a Cron expression to specify at which day(s) and time this event should trigger.
   For now only specify the event to start servers (E.g. at the beginning of the weekday). We will confire the second event later in this tutorial.
   E.g. cron(00 06 ? * MON-FRI *) fires every weekday (Monday to Friday) at 6:00 AM UTC. See this website for a handy Cron calculator.
8. Enable the Enable trigger checkbox.
9. Click Next.

Configure function

10. On the Configure function page, enter a Name and Description for the function.
11. For Runtime, select Python 2.7.
12. For Code entry type make sure Edit code inline is selected.
13. Delete the existing code and paste the following code into the code field:

import os
import boto3

client = boto3.client('autoscaling')

def get_env_variable(var_name):
    msg = "Set the %s environment variable"
    try:
        return os.environ[var_name]
    except KeyError:
        error_msg = msg % var_name

def lambda_handler(event, context):
    auto_scaling_groups = get_env_variable('NAMES').split()

    for group in auto_scaling_groups:
        if servers_need_to_be_started(group):
            action = "Starting"
            min_size = int(get_env_variable('MIN_SIZE'))
            max_size = int(get_env_variable('MAX_SIZE'))
            desired_capacity = int(get_env_variable('DESIRED_CAPACITY'))
        else:
            action = "Stopping"
            min_size = 0
            max_size = 0
            desired_capacity = 0

        print action + ": " + group
        response = client.update_auto_scaling_group(
            AutoScalingGroupName=group,
            MinSize=min_size,
            MaxSize=max_size,
            DesiredCapacity=desired_capacity,
        )

        print response

def servers_need_to_be_started(group_name):
    min_group_size = get_current_min_group_size(group_name)
    if min_group_size == 0:
        return True
    else:
        return False
    

def get_current_min_group_size(group_name):
    response = client.describe_auto_scaling_groups(
        AutoScalingGroupNames=[ group_name ],
    )
    return response["AutoScalingGroups"][0]["MinSize"]

14. For Environment variables add the following:
    NAMES - Space separated list of the Auto Scaling Groups you want to manage with this function
    MIN_SIZE - Minimum size of the Auto Scaling Group(s) when EC2 instances are started
    MAX_SIZE - Maximum size of the Auto Scaling Group(s) when EC2 instances are started
    DESIRED_CAPACITY - Desired capacity of the Auto Scaling Group(s) when EC2 instances are started

15. For Handler, make sure the value is lambda_function.lambda_handler.

16. For Role, select Create a custom role. An IAM wizard will open in a new tab.

17. For IAM Role, select Create a new IAM Role.

18. Enter a Role Name.

19. Click on View Policy Document and click Edit.

20. A warning popup will appear. Click Ok.

21. Add the following statement right after the closing } of "Resource": "arn:aws:logs:*:*:*" }

,
{
      "Effect": "Allow",
      "Action": "autoscaling:*",
      "Resource": "*"
    }

22. Click on Allow.

23. Back on the Configure function screen, make sure your new Role is selected under Existing role.

24. Click on Next and Create Function.

Create Stop event

25. To configure the second event (to stop your servers at the end of the weekday), on the Triggers tab of your new Lambda function, click on Add trigger.
26. In the Add trigger popup, from the Rule dropdown, select Create a new rule.
27. Enter a Rule name and Rule description.
28. For Rule type, select Schedule expression.
29. For Schedule expression, enter a Cron expression to specify at which day(s) and time this event should trigger.
    E.g. cron(00 20 ? * MON-FRI *) fires every weekday (Monday to Friday) at 20:00 PM UTC. See this website for a handy Cron calculator.
30. Enable the Enable trigger checkbox.

That's it, all done!
You can test the function by hitting the Test button. The first time an Input test event popup will appear.
For Sample event template select Scheduled event and click Save and test.

Acknowledgement

This tutorial on Nick Todd's blog helped me setting up this tutorial.