start-server-and-test

What is start-server-and-test?

The start-server-and-test npm package is designed to simplify the process of starting a server, running tests, and then shutting down the server. It is particularly useful in continuous integration environments where you need to ensure that the server starts successfully before tests are executed and is properly shut down afterward. This package helps automate these steps by providing a single command that handles the entire process.

What are start-server-and-test's main functionalities?

Start server, run tests, and stop server

This feature allows you to start your server using a start command (e.g., 'npm run start'), wait for the server to be accessible on a specified URL (e.g., http://localhost:3000), and then run tests using a test command (e.g., 'npm run test'). After the tests are completed, the server is automatically stopped.

start-server-and-test 'npm run start' http://localhost:3000 'npm run test'

Other packages similar to start-server-and-test

concurrently

Concurrently is a package that allows you to run multiple commands concurrently. Unlike start-server-and-test, it does not specifically manage the lifecycle of starting a server, running tests, and then stopping the server, but it can be used to achieve similar functionality with custom scripts.

forever

Forever is a simple CLI tool for ensuring that a given script runs continuously (i.e., forever). It is different from start-server-and-test as it focuses on keeping a script running rather than managing the start-test-stop cycle, but it can be used in combination with other tools to achieve similar end results.

pm2

PM2 is a powerful, production process manager for Node.js applications with a built-in load balancer. It is more complex and feature-rich compared to start-server-and-test, which is specifically focused on integration testing environments rather than full application management.

README

start-server-and-test

Starts server, waits for URL, then runs test command; when the tests end, shuts down server

Install

Requires Node version 8.9 or above.

npm install --save-dev start-server-and-test

Use

This command is meant to be used with NPM script commands. If you have a "start server", and "test" script names for example, you can start the server, wait for a url to respond, then run tests. When the test process exits, the server is shut down.

{
    "scripts": {
        "start-server": "npm start",
        "test": "mocha e2e-spec.js",
        "ci": "start-server-and-test start-server http://localhost:8080 test"
    }
}

To execute all tests simply run npm run ci.

Commands

In addition to using NPM script names, you can pass entire commands (surround them with quotes so it is still a single string) that will be executed "as is". For example, to start globally installed http-server before running and recording Cypress.io tests you can use

# run http-server, then when port 8000 responds run Cypress tests
start-server-and-test 'http-server -c-1 --silent' 8000 './node_modules/.bin/cypress run --record'

Because npm scripts execute with ./node_modules/.bin in the $PATH, you can mix global and locally installed tools when using commands inside package.json file. For example, if you want to run a single spec file:

{
  "scripts": {
    "ci": "start-server-and-test 'http-server -c-1 --silent' 8080 'cypress run --spec cypress/integration/location.spec.js'"
  }
}

Or you can move http-server part into its own start script, which is used by default and have the equivalent JSON

{
  "scripts": {
    "start": "http-server -c-1 --silent",
    "ci": "start-server-and-test 8080 'cypress run --spec cypress/integration/location.spec.js'"
  }
}

Here is another example that uses Mocha

{
  "scripts": {
    "ci": "start-server-and-test 'http-server -c-1 --silent' 8080 'mocha e2e-spec.js'"
  }
}

Alias

You can use either start-server-and-test, server-test or start-test commands in your scripts.

You can use : in front of port number like server-test :8080, so all these are equivalent

start-server-and-test start http://localhost:8080 test
server-test start http://localhost:8080 test
server-test http://localhost:8080 test
start-test :8080 test
start-test 8080 test
start-test 8080

Options

If you use convention and name your scripts "start" and "test" you can simply provide URL

{
    "scripts": {
        "start": "npm start",
        "test": "mocha e2e-spec.js",
        "ci": "start-server-and-test http://localhost:8080"
    }
}

You can also shorten local url to just port, the code below is equivalent to checking http://localhost:8080.

{
    "scripts": {
        "start": "npm start",
        "test": "mocha e2e-spec.js",
        "ci": "server-test 8080"
    }
}

You can provide first start command, port (or url) and implicit test command

{
    "scripts": {
        "start-it": "npm start",
        "test": "mocha e2e-spec.js",
        "ci": "server-test start-it 8080"
    }
}

You can provide port number and custom test command, in that case npm start is assumed to start the server.

{
    "scripts": {
        "start": "npm start",
        "test-it": "mocha e2e-spec.js",
        "ci": "server-test :9000 test-it"
    }
}

You can provide multiple resources to wait on, separated by a pipe |. (be sure to wrap in quotes)

{
  "scripts": {
    "start": "npm start",
    "test-it": "mocha e2e-spec.js",
    "ci": "server-test \"8080|http://foo.com\""
  }
}

or for multiple ports simply: server-test '8000|9000' test.

If you want to start the server, wait for it to respond, and then run multiple test commands (and stop the server after they finish), you should be able to use && to separate the test commands:

{
  "scripts": {
    "start": "npm start",
    "test:unit": "mocha test.js",
    "test:e2e": "mocha e2e.js",
    "ci": "start-test 9000 'npm run test:unit && npm run test:e2e'"
  }
}

The above script ci after the localhost:9000 responds executes the npm run test:unit command. Then when it finishes it runs npm run test:e2e. If the first or second command fails, the ci script fails. Of course, your mileage on Windows might vary.

expected

The server might respond, but require authorization, returning an error HTTP code by default. You can still know that the server is responding by using --expect argument (or its alias --expected):

$ start-test --expect 403 start :9000 test:e2e

See demo-expect-403 NPM script.

npx and yarn

If you have npx available, you can execute locally installed tools from the shell. For example, if the package.json has the following local tools:

{
  "devDependencies": {
    "cypress": "3.2.0",
    "http-server": "0.11.1",
    "start-server-and-test": "1.9.0"
  }
}

Then you can execute tests simply:

$ npx start-test 'http-server -c-1 .' 8080 'cypress run'
starting server using command "http-server -c-1 ."
and when url "http://localhost:8080" is responding
running tests using command "cypress run"
Starting up http-server, serving .
...

Similarly, you can use yarn to call locally installed tools

$ yarn start-test 'http-server -c-1 .' 8080 'cypress run'
yarn run v1.13.0
$ /private/tmp/test-t/node_modules/.bin/start-test 'http-server -c-1 .' 8080 'cypress run'
starting server using command "http-server -c-1 ."
and when url "http://localhost:8080" is responding
running tests using command "cypress run"
Starting up http-server, serving .
...

Note for yarn users

By default, npm is used to run scripts, however you can specify that yarn is used as follows:

"scripts": {
    "start-server": "yarn start",
    "test": "mocha e2e-spec.js",
    "ci": "start-server-and-test 'yarn start-server' http://localhost:8080 'yarn test'"
}

Note for webpack-dev-server users

Also applies to Vite users!

If you are using webpack-dev-server (directly or via angular/cli or other boilerplates) then the server does not respond to HEAD requests from start-server-and-test. You can check if the server responds to the HEAD requests by starting the server and pinging it from another terminal using curl

# from the first terminal start the server
$ npm start
# from the second terminal call the server with HEAD request
$ curl --head http://localhost:3000

If the server responds with 404, then it does not handle the HEAD requests. You have two solutions:

Use HTTP GET requests

You can force the start-server-and-test to ping the server using GET requests using the http-get:// prefix:

start-server-and-test http-get://localhost:8080

Ping a specific resource

As an alternative to using GET method to request the root page, you can try pinging a specific resource, see the discussion in the issue #4.

# maybe the server responds to HEAD requests to the HTML page
start-server-and-test http://localhost:3000/index.html
# or maybe the server responds to HEAD requests to JS resource
start-server-and-test http://localhost:8080/app.js

Explanation

You can watch the explanation in the video Debug a Problem in start-server-and-test.

Under the hood this module uses wait-on to ping the server. Wait-on uses HEAD by default, but webpack-dev-server does not respond to HEAD only to GET requests. Thus you need to use http-get:// URL format to force wait-on to use GET probe or ask for a particular resource.

Debugging

To see diagnostic messages, run with environment variable DEBUG=start-server-and-test

$ DEBUG=start-server-and-test npm run test
  start-server-and-test parsing CLI arguments: [ 'dev', '3000', 'subtask' ] +0ms
  start-server-and-test parsed args: { services: [ { start: 'npm run dev', url: [Array] } ], test: 'npm run subtask' }
...
making HTTP(S) head request to  url:http://localhost:3000 ...
  HTTP(S) error for http://localhost:3000 Error: Request failed with status code 404

Disable HTTPS certificate checks

To disable HTTPS checks for wait-on, run with environment variable START_SERVER_AND_TEST_INSECURE=1.

Timeout

This utility will wait for maximum of 5 minutes while checking for the server to respond (default). Setting an environment variable WAIT_ON_TIMEOUT=600000 (milliseconds) sets the timeout for example to 10 minutes.

Interval

This utility will check for a server response every two seconds (default). Setting an environment variable WAIT_ON_INTERVAL=600000 (milliseconds) sets the interval for example to 10 minutes.

Starting two servers

Sometimes you need to start one API server and one webserver in order to test the application. Use the syntax:

start-test <first command> <first resource> <second command> <second resource> <test command>

For example if API runs at port 3000 and server runs at port 8080:

{
  "scripts": {
    "test": "node src/test",
    "start:api": "node src/api",
    "start:server": "node src/server",
    "test:all": "start-test start:api 3000 start:server 8080 test"
  }
}

In the above example you would run npm run test:all to start the API first, then when it responds, start the server, and when the server is responding, it would run the tests. After the tests finish, it will shut down both servers. See the repo start-two-servers-example for full example

Note for Apollo Server users

When passing a simple GET request to Apollo Server it will respond with a 405 error. To get around this problem you need to pass a valid GraphQL query into the query parameter. Passing in a basic schema introspection query will work to determine the presence of an Apollo Server. You can configure your npm script like so:

{
  "scripts": {
    "ci": "start-server-and-test start 'http-get://localhost:4000/graphql?query={ __schema { queryType { name } } }' test"
  }
}

Small print

Author: Gleb Bahmutov <[e-mail]> © 2017

• @bahmutov
• glebbahmutov.com
• blog

License: MIT - do anything with the code, but don't blame me if it does not work.

Support: if you find any problems with this module, email / tweet / open issue on Github

MIT License

See LICENSE