About Imposters and Doubles

Photo by wild vibez on Unsplash

Well into the project, with the deadline looming, in the middle of the sprint, and you have half or more of your team browsing the internet because some upstream system is down and no development/testing/showcase can be done.

If you are part of a high performant team, that have all their microservices architecture and pipelines working this kind of problem probably does not occur to you.

But to most of us it does.

All the time.

You may actually have already tried stubbing(or mocking) some or all of your upstream endpoints but there is always a case that you haven’t thought or it is too hard to recreate with a stub (or a mock).

(In case you did not understand, or are vaguely familiar with, the last paragraph, stubs or mocks are simple versions of your complex backends that allow your team to keep working towards the project goals even when your backend is down.)

Mocks and stubs, more generally: test doubles, also are very useful in the early stages of the project when you have a team of developers but neither the requirements nor the design are yet complete and you don’t want to waste their valuable hourly rates but instead let them help in the initial contracts of the application.

So a very common approach is to create dummy servers that respond to specific requests. For example an end point that returns with a fixed address that can act as a double for an address lookup in your form.

This traditional approach involves some work with a stubbing framework or a simple webserver for which a developer prepares a suite of fixed responses.

A common tool that allows this is stubby4j. This is a standalone jar (with Jetty inside) that allows to write your responses to your mock endpoints in yml.

This can be a good start, however each response needs to be crafted carefully by hand, limiting the ability to produce realistic responses, only simple ones. So most likely this method won’t scale with your project development lifecycle.

Enter Mountebank.

Mountebank is a nice tool that allows to create your responses proxying your real backend and then saving all the responses for automatic replaying.

So basically you tell Mountebank the address of your backend and point your application to the Mountebank address. Then ask your testers/business analysts/developers to use your application as per normal, and then you can switch your backend off, creating an imposter.

The more you use Mountebank as a proxy to your application the better your Imposter will be.

Forget about asking a developer to craft manually hundreds of lines of potential responses or worse multiline RegExps that nobody would like to maintain.

Not only that, but Mountebank allows you to record all these imposters as a config file and you can start it anytime with this set of preconfigured doubles.

How to use it then?

The documentation on their site is very nice, but it assumes a lot of knowledge, so for the remainder of this article, I will show how to set it up with a sample API and also compare it with stubby4j.

Photo by Shane Rounce on Unsplash

Hands on stuff

To continue with the article clone/fork the following repo: https://github.com/momenton/mountebank-demo

You will also need:

• Mountebank (download from: http://www.mbtest.org/docs/install the option for Self-contained archives choosing your platform from the list)
• Docker (https://www.docker.com/)
• cURL installed in your terminal
• Java (optional, for stubby4j only)
• NodeJS (optional, if you don’t want to use Docker)
• Postman (optional)

Assuming you have cloned the repo to the following location:

runnerdave@runnerdave-N56VM:~/projects/mountebank-demo$

and downloaded and expanded Mountebank to this location (linux example):

runnerdave@runnerdave-N56VM:~/dev/javascript/mountebank/mountebank-v1.14.1-linux-x64$

Then from your mountebank-demo location run docker-compose:

And start Mountebank:

Now you are ready to create your first Impostor: (copy and paste into your terminal)

curl -i -X POST -H 'Content-Type: application/json' http://127.0.0.1:2525/imposters --data '{
  "port": 6569,
  "protocol": "http",
  "name": "proxyOnce",
  "stubs": [
    {
      "responses": [
        {
          "proxy": {
            "to": "http://localhost:8125",
            "mode": "proxyOnce",
            "predicateGenerators": [
              {
                "matches": {
                  "method": true,
                  "path": true,
                  "query": true,
                  "body": true
                }
              }
            ]
          }
        }
      ]
    }
  ]
}'

then hit the following endpoint in your browser: http://localhost:6569/api/customers

and compare it with the api response directly: http://localhost:8125/api/customers

you can notice they are equivalent.

Now you can see your imposters here:
http://localhost:2525/imposters/6569

you should see the following:

Then hit the following POST endpoint via cURL (or Postman):

curl -X POST \
  http://localhost:6569/api/create-customer \
  -H 'Cache-Control: no-cache' \
  -H 'Content-Type: application/json'
  -d '{"firstname": "Maria","lastname": "Sharapova"}'

This would have created another imposter that you can inspect using the same url as before.

And here is where the magic happens: stop your docker container ([ctrl]+c), and hit your previous endpoints again, and they should reply as before.

You can also save this run with the following call (ensure mountebank is still running):

.../mountebank-v1.14.1-linux-x64$ ./mb save --savefile impostorsForSimplestPossibleAPI.json

You can stop Mountebank now (([ctrl]+c)) and start it with the reference to the previous file:

.../mountebank-v1.14.1-linux-x64$ ./mb --configfile impostorsForSimplestPossibleAPI.json

Further reading and references

• Martin Fowler, Mocks aren’t stubs
• Otávio Machado, Using Mountebank to stub external dependencies
• Mountebank stubs (http://www.mbtest.org/docs/api/stubs)

If you would like to learn more head on over to momenton and hit us up for a chat.