Serverless boilerplate to Google-functions

This is a boilerplate for the Serverless framework, and google as a provider, tests with Jest, configured Eslint, airbnb standard, sucrase to support imports.

Tech Stack:

Sucrase
Eslint + Prettier (Airbnb config)
Tests with Jest
Nodemon
Serverless framework
Google-cloud/functions-emulator / firebase-tools

How to setup, write, test and deploy your own serverless functions

Prepare your local machine

Step 1

Google projects

We have many GCP projects. Here is the list of them. Developers should use playground for all of their testing. You should not create new projects. Prod, QA and Dev have deployment pipelines to deploy cloud functions.

Playground (strong-keyword-123)

experimental environment for all service proof of concepts

Platform-Dev (platform-dev-1234)

development environment for automated deployments and initial testing

Platform-QA (platform-qa-12345)

Qa environment for integration tests

Platform-Prod (platform-prod-123456)

Production environment

Step 2

Install gcloud cli

https://cloud.google.com/sdk/downloads

Once the tar.gz is installed initialize your gcloud...

gcloud init

Choose account to perform operations for this configuration

Select your domain name

Pick cloud project to use

Select project for playground (strong-keyword-123)

Do you want to configure a default Compute Region and Zone?

Answer Yes

Set region zone to (us-central1)

Update components and enable beta

gcloud components update && gcloud components install beta

Init again to edit these settings or add other configurations

Also you can override your project and zone with the --project and --zone options

Example to list functions from non-default project and zone gcloud beta functions list --project platform-dev-1234 --zone us-central1-f

Step 3

Setup node.js

Install nvm

https://github.com/creationix/nvm#installation

Google Cloud Functions support node.js v6.11.5 and v8.15.0

Install node.js v8.15.0

nvm install v8.15.0

nvm use v8.15.0

Step 4

Install serverless framework

https://serverless.com/

Install with npm

npm install serverless -g

This will install serverless into the nvm env, if you wish to have it available in your standard path

ln -fs /Users/{user_name}/.nvm/versions/node/v{node_version}/bin/serverless /usr/local/bin

Run the tool

serverless --version

Serverless authentication

A keyfile.json is necessary to deploy serverless functions in our playground project

Go to the API Credentials page in GCP

https://console.cloud.google.com/apis/credentials?project=strong-keyword-1234&organizationId=123456

Select Create Credentials

Select Service Account Key

Under Service Account select the “serverless” service account

Ensure JSON is selected

Press Create

A json file will download, move it to ~/.gcloud/keyfile.json

Step 5

Install cloud-function-emulator

https://cloud.google.com/functions/docs/emulator

Install with npm

npm install -g @google-cloud/functions-emulator

Run the tool (some shells will have a conflict with the default functions, use functions-emulator)

functions-emulator --help

Creating a serverless project

When starting a new serverless project first consider what feature or service your functions will serve. Check to see if a serverless project already exists for that feature or service. If not create a new project using the ORM project template.

Create a function from the ORM template Pull down the ORM cloud function template and give it a unique named directory serverless install --url https://github.com/luizfelipeneves/serverless-google-function-template --name {name_of_your_serverless_project}

You now have the ORM default template for google cloud functions installed in your directory. This includes a .gitignore, .eslintrc, .gcloudignore, README.md, index.js, package.json, serverless.yml, as well as a tests and functions directories.

Update package.json including name, version, description, tests, and dependencies

Update serverless.yml including service name, function name(s) and provider project fields.

Update README.md to describe the service your functions will serve.

Once set git init and create a repo in git and push your initial commit.

Developing a function

During the development lifecycle we are using node 6.11.5 and es6 as standards for writing. ESlint and jest are standards for linting and testing.

Notice in the project template there is a a single index.js module, this stores the set of functions that serve a similar feature or service. By default there are 2 sample functions httpHelloWorld and eventHelloWorld. http* is triggered by a web request while event is triggered by pubsub events.

Cloud functions are limited to deploying functions from one module so when writing if multiple function deployments will reside in one project they must be exported from the main index.js. AWS Lambda and Kubeless do not have this limitation and multiple js modules can be deployed simultaneously. This is a current limitation with google cloud functions only

We can use google-cloud-emulator to deploy and test a function locally to ensure it is working as expected. Serverless offers this with their invoke local method but it does not support google cloud functions… yet.

Start the emulator

functions-emulator start

Deploy the exported function name with an http trigger (as opposed to pub/sub event trigger)

functions-emulator deploy {my_function} --trigger-http

Call the deployed function

functions-emulator call {my_function}

Inspect the function from a browser

functions-emulator inspect {my_function}

View logs to get the Debugger url

functions_emulator logs read

You will see output like... Debugger listening on ws://127.0.0.1:9229/fb9906cd-a063-42ba-bcf2-af04fc2e2655

Copy and paste the url to your browser and call the function. You can now see vars, step through code and inspect your function to the fullest.

Turn off inspection

functions_emulator reset {my_function}

Stop the emulator

functions-emulator stop

Deploying a function

Deploy your function As you prepare to deploy functions be sure you are only deploying to playground from your local computer. Dev, QA and Prod will have deployments based on git hooks.

Serverless framework can deploy functions to gcp, aws, kubeless, and many more. With a simple change of the serverless.yml this function can be running in multiple clouds.

First package your functions to ensure they compile and package properly, find the results in the .serverless directory

serverless package

Once your function is tested with functions-emulator and receive the proper output you can push it to the serverless provider set in your serverless.yml

serverless deploy

Once deployed you can now get back logs from your function

serverless logs -f my_function

Serverless Plugins: Serverless framework allows us to use, or create, plugins to facilitate certain tasks.

By default our cloud-function-template uses serverless-google-cloudfunctions, but there are others that may be beneficial. To use a plugin they must be installed in the package.json AND referenced in the serverless.yml file plugins.

List available plugins

serverless plugins list

LuizFelipeNeves / Serverless-google-function-template