Lambda and Python

Using AWS Lambda to run Python scripts

James Mitchell

## Deploying Your App What needs to be deployed... ![Deployment Matrix](talk/aws-lambda-deployment-matrix.png)
## What is AWS Lambda? "_AWS Lambda is a compute service that runs your code in response to events and automatically manages the compute resources for you..._" * deploy __functions__ not applications * the code is executed in __response to events__ on other AWS services, such as S3, SNS, DynamoDB etc.. * or executed by the Amazon __API Gateway__ (⇦ choose this one!) * only supports Java and Node.js ☹
## Adding some Python We can make it work by getting a node.js script to execute the Python script! * [WillyG - Python on AWS Lambda](http://willyg302.github.io/blog/posts/2015-03-29-python-on-aws-lambda/) * [Tim Wagner - Using Python in an AWS Lambda Function](https://aws.amazon.com/blogs/compute/using-python-in-an-aws-lambda-function/) * [Eric Hammond provides a wrapper function](https://github.com/alestic/lambda-function-wrapper)
## 1.1 - _Hello World_ in Python * accept a JSON object from the wrapper, write JSON to stdout * use virtualenv to bundle in python and extra libraries 

import sys
import json

def main(event):
    name = event.get('name', 'Mr. Eastwood')
    response = dict(greeting='Hello', name=name)
    print(json.dumps(response))

if __name__=='__main__':
    argv = sys.argv[1:]
    event = json.loads(argv[0])
    main(event)
[sample code on Github](https://github.com/jtmitchell/lambda-function-wrapper)
## 1.2 - Virtualenv We need to create the virtualenv **inside** the project, so it can be included in the zipfile. 

project folder
  ├── lambda-function.py
  ├── lambda-function-wrapper.js
  ├── requirements.txt
  └── venv
      ├── bin
...
      │   ├── python -> python3.4
      │   ├── python3 -> python3.4
      │   └── python3.4
      ├── include
      │   └── python3.4m -> /usr/include/python3.4m
      └── lib
          └── python3.4
              ├── abc.py -> /usr/lib/python3.4/abc.py
              ├── base64.py -> /usr/lib/python3.4/base64.py
              ├── bisect.py -> /usr/lib/python3.4/bisect.py
...
## 2.1 - Lambda Wrapper Script * This script runs our Python script * Communication is via "stringified" JSON 

var spawn = require('child_process').spawn;
exports.handler = function(event, context) {
  var response = {};
  var error = null;
  var child = spawn('venv/bin/python', [
    'lambda-function.py',
    JSON.stringify(event, null, 2)
    ]);
  child.stdout.on('data', function (data) {
    console.log("stdout:\n"+data);
    response = JSON.parse(data);
  });
  child.stderr.on('data', function (data) {
    console.log("stderr:\n"+data);
    error = { error: true, message: data.toString('utf8') };
  });
  child.on('close', function (code) {
    if (error !== null ) { context.fail(error); }
    else {context.succeed(response); }
  });
};
## 2.1 - Lambda Function * upload a ZIP of the directory to S3 bucket * ```zip -r /tmp/hello-lambda.zip *``` * upload... this takes a while to upload 39Mb of Python * make an IAM role for executing the function * create the Lambda function 

		$ aws lambda create-function \
--region us-east-1 \
--function-name helloworld \
--code S3Bucket=maungawhau.lambda,S3Key=hello-lambda.zip \
--role arn:aws:iam::569584872835:role/lambda_basic_execution \
--handler lambda-function-wrapper.handler \
--runtime nodejs
## 2.3 - AWS Lambda Console Once it is running, use the console to check logs and number of calls. [![AWS Lambda Console](talk/aws-lambda-console-small.png)](talk/aws-lambda-console.png)
## 5.1 - API Gateway Configure a new API on AWS, with a resource that executes the Lambda function. [![AWS API Gateway](talk/aws-gateway-1-small.png)](talk/aws-gateway-1.png)
## 5.2 - Say “Hello” to my little friend 

$ curl -X GET https://hostname/test/greeting
{"name":"Mr. Eastwood","greeting":"Hello"}

$ curl -X POST -d '{"name": "Clint"}' https://hostname/test/greeting
{"name":"Clint","greeting":"Hello"}
## 5.3 - Again... with Pictures [![AWS API Gateway](talk/aws-gateway-4-small.png)](talk/aws-gateway-4.png)
## Conclusions * it's __cheap__ * FREE - first million requests per month * then $0.20 per 1 million requests * minimal deployment hassle * no EC2 instance, no server maintenance * great for a __micro-service__ * simple endpoint * reacting to an event * terrible for a Django app