Skip to content

Latest commit

 

History

History
530 lines (363 loc) · 17.9 KB

README.md

File metadata and controls

530 lines (363 loc) · 17.9 KB

RailsConf 2023: Webhooks Workshop

Catch Me If You Can: Learning to Process Webhooks in Your Rails App

This workshop will be taught on on Tuesday, April 25th at RailsConf 2023 in Atlanta.

In this workshop, you’ll learn how to catch and process webhooks like a pro. You’ll build a Rails app that’s both robust and low-latency so you keep up in real-time like a champ. Come ready to level up your skills and leave with the expertise you need to become a true webhook wizard!

We will begin by exploring the fundamentals of webhooks: how they work, why they are useful, and how they differ from other approaches. We will then dive into the hard-won lessons learned for consuming and processing webhooks, including routing, handling payloads, and responding to events.

Along the way, we will explore best practices like error handling, authentication, retries, idempotency, and scaling. You’ll walk away with a solid understanding for how to build a resilient and robust system to handle webhook notifications from a wide range of external services and APIs. Attendees will leave the workshop with a working webhook processor running on Rails!

How To Use This Repository

There are many ways to support webhooks in your application. For the interest of this workshop, we will be covering a straightforward approach that uses common Rails patterns from end to end.

As you evolve your own webhook handling and develop your own style, you will find areas that might benefit from middleware and more intermediate/advanced methods.

Branches

To allow you to follow along, we've set up each step as a branch so you can see the difference from each step. main contains the final project with all the PRs merged but you can checkout the start-here branch to move to the starting point. We have included instructions below under Getting Started.

You can check out each subsequent step with the following branches:

  • git checkout start-here
  • git checkout step1-routes
  • git checkout step2-webhook-model
  • git checkout step3-background-job
  • git checkout step4-verification
  • git checkout step5-tests

Webhooks Tutorial

Star this repository so we know you have taken this workshop!

Pull down the workshop repository

Pull down the repository and switch to the start-here remote branch

# clone from Github
git clone git@github.com:colinloretz/railsconf-webhooks.git

# move into directory
cd railsconf-webhooks

# fetch all the upstream branches that include the steps of this workshop
git fetch --all

# checkout the start-here branch
git checkout start-here

# install dependencies
bundle install

# start server
rails server

You should now have a brand new Rails app up and running. You can verify by visiting http://localhost:3000

Step 1: Setting Up a Controller and Rails Routes

We'll start by creating a controller and an HTTP route to catch the webhook events.

1A) Creating a Base Controller

We'll start by creating a base controller that all of our webhook controllers will inherit from. This will allow us to add common functionality to all of our webhook controllers.

rails generate controller Webhooks::BaseController
# app/controllers/webhooks/base_controller.rb

class Webhooks::BaseController < ApplicationController
  # Disable CSRF checking on webhooks because they do not originate from the browser
  skip_before_action :verify_authenticity_token

  def create
    head :ok
  end
end

For this workshop we are going to catch webhooks from two sources: a fake webhook provider Movies and Stripe. We'll create a controller for each of these webhook sources and inherit the Webhook::BaseController

1B) Creating the Movies Webhook Controller

This fake Movies controller also includes an example of how to send a webhook to your application. You can use this to test your webhook processor.

rails generate controller Webhooks::MoviesController
# app/controllers/webhooks/movies_controller.rb

class Webhooks::MoviesController < Webhooks::BaseController
  # A controller for catching new movie webhooks
  #
  # To send a sample webhook locally:
  #
  #   curl -X POST http://localhost:3000/webhooks/movies
  #     -H 'Content-Type: application/json'
  #     -d '{"title":"Dungeons & Dragons: Honor Among Thieves"}'
  #
  # If you'd like to override the base controller's behavior, you can do so here
  # def create
  #   head :ok
  # end
end

Let's create the Stripe webhook controller while we are here.

1C) Creating the Stripe Webhook Controller

rails generate controller Webhooks::StripeController
# app/controllers/webhooks/stripe_controller.rb

class Webhooks::StripeController < Webhooks::BaseController
  # If you'd like to override the base controller's behavior, you can do so here
  # def create
  #   head :ok
  # end
end

1D) Adding Routes for the Webhook Controllers

Finally, we need to add routes for these webhook controllers.

We are going to set up routes that are namespaced to /webhooks and then have a route for each webhook controller. We will only allow the create action on these routes.

As you add more webhook providers, you can also create a dynamic route that will catch all webhooks and route them to a single controller. This is a common pattern for webhook processors.

# config/routes.rb

Rails.application.routes.draw do
  namespace :webhooks do
    # /webhooks/movies routed to our Webhooks::MoviesController
    resource :movies, controller: :movies, only: [:create]

    # /webhooks/stripe routed to our Webhooks::StripeController
    resource :stripe, controller: :stripe, only: [:create]
  end

  # your other routes here
end

1E) Testing our new routes

We can test our new routes by sending a request to our new webhook routes. We can use curl to send a request to our new webhook routes.

Let's boot up the server and send a request to our new webhook routes.

This should be successful and return a 200 status code.

curl -X POST 'http://localhost:3000/webhooks/movies' -H 'Content-Type: application/json' -d '{"title":"Dungeons & Dragons: Honor Among Thieves"}' -v

We can't easily test our Stripe webhooks without setting up a Stripe account and configuring it to send webhooks to our application. We'll include instructions for this in a later step.

Now that we have our routes set up, we can move on to creating a webhook model.

💡 You can checkout the branch step1-routes using git checkout step1-routes to get caught up to this step before continuing.

Step 2: Creating a Webhook Model

This model will be responsible for storing the inbound webhook until a background job can process it.

The reason we want to do this is because we want our controller to respond as fast as possible for heavy traffic (like Black Friday for example) and we want to store the record in the database to safely store it for retries if the service is having trouble for any reason.

2A) Creating the Webhook Model

We can use the Rails generator to create our model and the associated helper files.

rails generate model InboundWebhook status:string body:text

We should now have a pretty empty model file and a migration file.

# app/models/inbound_webhook.rb

class InboundWebhook < ApplicationRecord
end
# db/migrate/20230401162628_create_inbound_webhooks.rb

class CreateInboundWebhooks < ActiveRecord::Migration[7.0]
  def change
    create_table :inbound_webhooks do |t|
      t.string :status, default: :pending
      t.text :body

      t.timestamps
    end
  end
end

We can run our migration to create the table in the database.

rails db:migrate

Now that we have a model and a backing table in the database, we can save our webhook to the database. Let's update our Webhook::BaseController or each of our webhook controllers to save the webhook to the database.

# app/controllers/webhooks/base_controller.rb

class Webhooks::BaseController < ApplicationController
  # Disable CSRF checking on webhooks because they do not originate from the browser
  skip_before_action :verify_authenticity_token

  def create
    InboundWebhook.create(body: payload)
    head :ok
  end

  private

  def payload
    @payload ||= request.body.read
  end
end

Now if you test your Movies webhook route, you should see a new record in the database.

curl -X POST 'http://localhost:3000/webhooks/movies' -H 'Content-Type: application/json' -d '{"foo":"bar"}'

💡 You can checkout the branch step2-webhook-model using git checkout step2-webhook-model to get caught up to this step before continuing.

Step 3: Creating a Background Worker

Now that we have a record in our database, we need to process it. We don't want to process it in the controller because that would slow down our response time. Instead, we want to process it in a background job.

3A) Creating the Background Job

Let's use another Rails generator to create a background job.

In thise case, we are going to use the ActiveJob framework that comes with Rails. This will allow us to easily switch between background job providers like Sidekiq, Resque, DelayedJob, etc.

Let's make a job for each of our webhook providers within a Webhooks namespace.

rails generate job Webhooks::MoviesJob
# app/jobs/webhooks/movies_job.rb

class Webhooks::MoviesJob < ApplicationJob
  queue_as :default

  def perform(inbound_webhook)
    webhook_payload = JSON.parse(inbound_webhook.body, symbolize_names: true)

    # do whatever you'd like here with your webhook payload
    # call another service, update a record, etc.

    inbound_webhook.update!(status: :processed)
  end
end
rails generate job Webhooks::StripeJob
# app/jobs/webhooks/stripe_job.rb

class Webhooks::StripeJob < ApplicationJob
  queue_as :default

  def perform(inbound_webhook)
    json = JSON.parse(inbound_webhook.body, symbolize_names: true)
    event = Stripe::Event.construct_from(json)
    case event.type
    when 'customer.updated'
      # Find customer and save changes
      inbound_webhook.update!(status: :processed)
    else
      inbound_webhook.update!(status: :skipped)
    end
  end
end

3B) Updating our Webhook Controllers to enqueue jobs

Now that we have our jobs defined, we need to enqueue them in our webhook controllers.

Let's update both of our controllers to process these webhooks by overriding the create method in both controllers.

# app/controllers/webhooks/movies_controller.rb

class Webhooks::MoviesController < Webhooks::BaseController
  def create
    # Save webhook to database
    record = InboundWebhook.create!(body: payload)

    # Queue database record for processing
    Webhooks::MoviesJob.perform_later(record)

    head :ok
  end

  private

  def payload
    @payload ||= request.body.read
  end
end

And we can set up our Stripe controller, similarly.

# app/controllers/webhooks/stripe_controller.rb

class Webhooks::StripeController < Webhooks::BaseController
  def create
    # Save webhook to database
    record = InboundWebhook.create!(body: payload)

    # Queue database record for processing
    Webhooks::StripeJob.perform_later(record)

    # Tell Stripe everything was successful
    head :ok
  end

  private

  def payload
    @payload ||= request.body.read
  end
end

💡 You can checkout the branch step3-background-job using git checkout step3-background-job to get caught up to this step before continuing.

Step 4: Verifying Webhooks

Verifying webhooks is an important step to make sure that the event is a real event that you expect from the Webhook Provider.

In our controllers, we have created a verify_event method that is called as a before_action in our base controller.

# app/controllers/webhooks/base_controller.rb

class Webhooks::BaseController < ApplicationController
  # Disable CSRF checking on webhooks because they do not originate from the browser
  skip_before_action :verify_authenticity_token

  before_action :verify_event

  def create
    InboundWebhook.create(body: payload)
    head :ok
  end

  private

  def verify_event
    head :bad_request
  end

  def payload
    @payload ||= request.body.read
  end
end

For testing purposes with our Movie webhooks, we have added a verify_event method that is always true unless you pass in a fail_verification url parameter on the webhook endpoint.

# app/controllers/webhooks/movies_controller.rb
# ... rest of controller

private

def verify_event
  head :bad_request if params[:fail_verification]
end

To verify Stripe webhooks in Webhooks::StripeController, Stripe provides a method in their ruby gem that uses a combination of a Webhook Signature, your Stripe signing secret and the JSON payload to verify the event.

# app/controllers/webhooks/stripe_controller.rb
# ... rest of controller

private

def verify_event
  signature = request.headers['Stripe-Signature']
  secret = Rails.application.credentials.dig(:stripe, :webhook_signing_secret)

  ::Stripe::Webhook::Signature.verify_header(
    payload,
    signature,
    secret.to_s,
    tolerance: Stripe::Webhook::DEFAULT_TOLERANCE,
  )
rescue ::Stripe::SignatureVerificationError
  head :bad_request
end

When consuming webhooks from other sources, you would customize verify_event to match the signature verification method that the Webhook Provider has in their webhook documentation.

For webhooks that don't provide verification methods, you can optionally call to the source API to check that the event has happened or that data has been created/updated as the event suggests.

However, it is recommended that you do this in the background job that processes the webhook so you don't make unnecessary API calls while returning a status back to the Webhook Provider.

💡 You can checkout the branch step4-verification using git checkout step4-verification to get caught up to this step before continuing.

Step 5: Writing tests for our Webhook Processor

Now that we have our webhook processor set up, we can write some tests to make sure that it works as expected.

We will start with a simple 200 OK status test on our MoviesController.

First, let's create a JSON file to represent our webhook. If you have a lot of webhooks, you can create a folder for all of your webhook fixtures and separate them by provider.

We'll start by creating a folder for our webhooks in test/fixtures and then create a JSON file for our Movie webhook.

mkdir test/fixtures/webhooks

And create a file for our Movie webhook.

touch test/fixtures/webhooks/movie.json
{
  "title": "Dungeons & Dragons: Honor Among Thieves",
  "release_date": "2023-03-31"
}
# spec/requests/webhooks/movies_controller_spec.rb
require 'test_helper'

class Webhooks::MoviesControllerTest < ActionDispatch::IntegrationTest
  def setup
    # Load the webhook data from the JSON file
    file_path = Rails.root.join('test', 'fixtures', 'webhooks', 'movie.json')
    @webhook = JSON.parse(File.read(file_path))
  end

  test 'should consume webhook' do
    # Send the POST request to the create action with the prepared data
    post webhooks_movies_url, params: @webhook

    # Check if the response status is 200 OK
    assert_response :ok

    # You can create other test files for each job or service that is called
    # For example, check if a record was created/updated in the database
  end
end

You should have a successful test run if you run the following command:

rails test

You can also use more advanced testing frameworks like Rspec and FactoryBot to write tests for your webhook processor.

💡 You can checkout the branch step5-tests using git checkout step5-tests to get caught up to this step before continuing.

Other Topics To Consider

Retries

Usually retries are done on the Webhook Producer side of things. However, if you are receiving a lot of webhooks and you are having trouble processing them all, you may want to implement background job retries in your application. When doing this, you will also want to make sure you have an idempotency method in place to ensure that you don't process the same webhook event more than once.

Preventing Replay Attacks: Idempotency & Deduplication

As a Webhook Consumer, you want to make sure that you don't process the same webhook event multiple times. This is especially important if you are doing something like updating a record in your database.

Most Webhook Providers offer a unique identifier or idempotency method for each webhook event. You can use this to ensure that you don't process the same webhook event more than once.

Backfilling Missing Events

The beauty of webhooks is that it is an evented-architecture. However, they are not always 100% reliable and a provider may not be able to deliver a webhook event to your application.

If a webhook provider is unable to deliver a webhook to your application, most will retry a certain number of times over a given time period (usually with exponential backoff). However, if after that time, they still were not able to deliver the webhook payload OR had some sort of service outage, you might be missing important webhook events.

You can implement other API calls to the service that runs on an interval or when other events are received to check if any other events happened since the last event was received.

Webhooks Best Practices & Tools

Popular Options for safe ingress into your application from the outside world