micro-di

readiness

A small utility module to check if a service is up and running at a given url.

Overview

Often when testing your code against dependent (micro)services, it's required to defer the test run until a given dependency is up and running - usually meaning it responds with a 2XX HTTP status-code at a dedicated health-check endpoint. Since services tend to have a relatively long starup time (imagine a JVM-based, dockerized service started as a test-dependency), it's often required to attempt the readiness call at given intervals for a certain period of time to allow the service to start-up and respond.

Using Readiness in tests

Whether you're using Jest or Mocha, using readiness looks quite the same. I'll use Jest to demonstrate:

const {readiness} = require('readiness')
 
describe('My code is awesome', () => {
   beforeAll(() => {
      await readiness('http://some-other-service/is_ready');
   });
 
   test('Other code - not so', () => {
      expect(fetch('http://localhost:3030/my-service')).resolves.toEqual(true));
   });
});

The call to readiness makes sure the the other service we depend on is ready (in the default use-case, responds with a 2XX status-code), and only then resolves the promise that signals beforeAll() to complete.

Options

readiness supports additional options that control the check procedure, and can be passed as a second argument to the readiness function:

readiness('http://localhost:3000', options)

1. options.attempts - the number of attempts to call the service before it's declared unavailble, rejecting the promise. Defaults to 1.
2. options.interval - the interval (in milliseconds) between consequent attempts to invoke the service. Effective only when options.attempts is greater than 1, and defaults to 200ms.
3. options.customCheck - a custom function that allows overriding the default readiness-check logic. Accepts the node-fetch response as an input, and expected to resolve if the response is considered to be from a healthy service, or reject otherwise. An example use-case is when you consider a service ready if it returns a 401 response code, since you cannot or does not want to authenticate against the health-check endpoint.

const {readiness} = require('readiness')
 
describe('My code is awesome', () => {
   beforeAll(() => {
      await readiness('http://some-other-service/is_ready', {
         customCheck: async (response) => {
            const isUnauthorized = response.status === 401;
            const body = await response.json();
            return isUnauthorized && body.error === 'unauthorized' ? Promise.resolve() : Promise.reject();
         }
      });
   });
 
   test('Other code - not so', () => {
      expect(fetch('http://localhost:3030/my-service')).resolves.toEqual(true)); // Succeeds, even though the service responded with a non-2XX code - due to the customCheck implementation
   });
});