instance_manager

Python script to start/stop instances in AWS, meant to be run as a Lambda function.

Uses the python-lambda toolset nficano/python-lambda .

Instructions

Setup

To setup, create a virtualenv and run pip install -r requirements.txt

AWS connection

To enable AWS connection you need to configure a default AWS profile in ~/.aws/credentials or provide keys in config.yaml.

Test run

Once the AWS connection has been configured you can test instance_manager from the local workstation. The file event.json contains sample data for test purposes.

{
  "tags": {
    "Purpose": "lambda-testing",
    "Testing": "true"
  },
  "state": "stop"
}

Basically the tags given here are matched against the EC2 instance tags and when a match is found the instance state is moved to one of start/stop/terminate/restart which ever is defined here.

Define the tags and state as you want and run the test with command lambda invoke -v.

Create IAM role for instance_manager

instance_manager needs certain access rights to be able to perform its tasks. The easiest way may be to create a new IAM role for it as follows.

IAM - Roles - Create New Role
Role name = lambda_instance_manager (defined in config.yaml)
Select Role Type - AWS Service Roles - AWS Lambda
Attach Policy (leave empty) - Next Step
Review - Create Role
Select lambda_instance_manager role - Permissions - Inline Policies - Create - Custom Policy - Select

Policy Name = InstanceManagerPolicy Policy Document

{
    "Version": "2012-10-17",
    "Statement": [
        {
            "Sid": "Stmt1474026789000",
            "Effect": "Allow",
            "Action": [
                "ec2:DescribeInstanceStatus",
                "ec2:DescribeInstances",
                "ec2:DescribeReservedInstances",
                "ec2:MonitorInstances",
                "ec2:StartInstances",
                "ec2:StopInstances"
            ],
            "Resource": [
                "*"
            ]
        }
    ]
}

Deploy instance_manager to AWS

To deploy instance_manager to AWS, just type lambda deploy. (make sure you're deploying to the correct account)

Once the deployment is complete verify that instance-manager (defined in config.yaml) is visible on the AWS Lambda Functions page.

Create CloudWatch schedule

The final step is to create the CloudWatch schedule to run the lambda.

CloudWatch - Rules - Create Rule
Event Source - Schedule - Cron Expression - Put your schedule here e.g. 30 15 ? * * * runs every day at 15:30 UTC
Add Target - Lambda Function - Function = instance-manager
Configure Input - Constant (JSON Text) - Paste your event.json here e.g. { "tags": { "Purpose": "lambda-testing", "Testing": "true" }, "state": "stop" }
Configure Details - Give name and description and click Create Rule

iluwatar/instance_manager