I am building a cdk stack including a bunch of lambda functions. My goal here is to update all the node to es6 so I can use tree shaking, and use Webapck to hugely optimise the individual lambda bundles. I’m also updating the AWS-SDK to version 3 so I can take advantage of the leaner imports of individual packages and use bare bones clients etc. Basically anything I can think of to trim any fat off the code so speed it up and save a few $$$.

The folder structure of the relevant parts of the project:

monkey-cloud/
├──lambda
│  ├── src/
│  │   ├── release/
│  │   │   ├── image-get.js
│  │   │   ├── image-put.js
│  │   ├── critical/
│  │   │   ├── data_iot.js
│  │   │   ├── data-get.js
│  │   │   ├── device-put.js
│  │   │   ├── device-get.js
│  ├── includes/
│  │   ├── dynamodb/
│  │   │   ├── monkey_dynamodb.js
│  │   ├── s3/
│  │   │   ├── monkey_s3.js
│  ├── dist/
├── webpack.config.js
├── package.json

What I have done so far is install webpack & a bunch of other stuff:

npm install --save-dev webpack webpack-cli babel-loader @babel/core @babel/preset-env

Then I’ve created a webpack.config.js at the root of the project:

const path = require('path');
const glob = require('glob');

// Automatically find all Lambda entry points
const entry = glob.sync('./lambda/src/**/*.js').reduce((acc, file) => {
    const name = path.relative('./lambda/src', file).replace(/.js$/, '');
    acc[name] = file;
    return acc;
}, {});

module.exports = {
    entry,
    output: {
        filename: '[name]/index.js',
        path: path.resolve(__dirname, 'lambda/dist'),
        libraryTarget: 'commonjs2',
    },
    mode: 'production',
    target: 'node',
    module: {
        rules: [
            {
                test: /.js$/,
                exclude: /node_modules/,
                use: {
                    loader: 'babel-loader',
                    options: {
                        presets: ['@babel/preset-env'],
                    },
                },
            },
        ],
    },
    optimization: {
        usedExports: true,
    },
};

I’ve added a simpole script into the package.json

"//-full-build": "webpack  && node ./lambda/lambda-zip.js",
"build": "webpack"

And last piece of the puzzle, I’ve added a little node.js script to to some build stuff (lambda-zip.js) – essentially to clean up the outputs into a format that is ready to use by cdk when I actually deploy the project. N.B. I’ve temporarily disabled that step above.

To test the progress so far, I can just run npm run build. Ideally, this should bundle up my lambdas into a nice clean tree of zipped files to shove into my cdk lambdas as code.

It’s not working so smoothly for me… When I run webpack, I end up with a lot of unexpected outputs. I’ve simplified the project to have only 1 lambda in it (critical/data-iot.js). When I just run webpack and nothing else, it does create the critical/data-iot.js in the dist folder. However, it builds a whole BUNCH of other folders called 109, 108 etc… And each of those has an index.js in them. It looks like it’s trying to build a bunch of lazy loaded assets…

Questions

1. How do I zip up this carnage into a single zip file for the cdk lambda?

2. When I add in the other hundred lambda functions, this is going to get slightly out of control!? Is there some way of forcing webpack to generate a single file for each lambda? And obviously in a sensible predetermined location. Otherwise it’s going to be a nightmare to automate the zipping up process.