Node.js Best Practices
Follow us on Twitter! @nodepractices
Read in a different language: 
CN, 
FR, 
BR, 
RU, 
PL, 
JA, 
EU (
ES, 
HE, 
KR and 
TR in progress! )
π We have an official Node.js starter - Practica.js. Use it to generate a new solution skeleton with all the practices baked inside. Or just it to learn by code examples
Latest Best Practices and News
-
β¨ 80,000 stars: Blushing, surprised and proud! -
π New menu and tags: Our menu is collapsible now and includes#tags. New visitors can read#strategicitems first. Returning visitors can focus on#newcontent. Seniors can filter for#advanceditems. Courtesy of the one and only Rubek Joshi -
π¨βπ©βπ§βπ¦ New family member!: A new repository joins our family - Node.js Integration Tests Best Practicesβ¨ . It includes 40+ best practices for writing awesome and performant Node.js component tests -
French translation!1! : The latest translation that joins our international guide is French. Bienvenue
Welcome! 3 Things You Ought To Know First
1. You are reading dozens of the best Node.js articles - this repository is a summary and curation of the top-ranked content on Node.js best practices, as well as content written here by collaborators
2. It is the largest compilation, and it is growing every week - currently, more than 80 best practices, style guides, and architectural tips are presented. New issues and pull requests are created every day to keep this live book updated. We'd love to see you contributing here, whether that is fixing code mistakes, helping with translations, or suggesting brilliant new ideas. See our writing guidelines here
3. Best practices have additional info - most bullets include a
Table of Contents
1. Project Structure Practices (5)
ββ1.1 Structure your solution by components #strategic
ββ1.2 Layer your components, keep the web layer within its boundaries #strategic
ββ1.3 Wrap common utilities as npm packages
ββ1.4 Separate Express 'app' and 'server'
ββ1.5 Use environment aware, secure and hierarchical config #modified-recently
2. Error Handling Practices (12)
ββ2.1 Use Async-Await or promises for async error handling
ββ2.2 Use only the built-in Error object #strategic
ββ2.3 Distinguish operational vs programmer errors #strategic
ββ2.4 Handle errors centrally, not within a middleware #strategic
ββ2.5 Document API errors using Swagger or GraphQL #modified-recently
ββ2.6 Exit the process gracefully when a stranger comes to town #strategic
ββ2.7 Use a mature logger to increase error visibility
ββ2.8 Test error flows using your favorite test framework
ββ2.9 Discover errors and downtime using APM products
ββ2.10 Catch unhandled promise rejections #modified-recently
ββ2.11 Fail fast, validate arguments using a dedicated library
ββ2.12 Always await promises before returning to avoid a partial stacktrace #new
3. Code Style Practices (12)
ββ3.1 Use ESLint #strategic
ββ3.2 Node.js specific plugins
ββ3.3 Start a Codeblock's Curly Braces on the Same Line
ββ3.4 Separate your statements properly
ββ3.5 Name your functions
ββ3.6 Use naming conventions for variables, constants, functions and classes
ββ3.7 Prefer const over let. Ditch the var
ββ3.8 Require modules first, not inside functions
ββ3.9 Require modules by folders, as opposed to the files directly
ββ3.10 Use the === operator
ββ3.11 Use Async Await, avoid callbacks #strategic
ββ3.12 Use arrow function expressions (=>)
4. Testing And Overall Quality Practices (13)
ββ4.1 At the very least, write API (component) testing #strategic
ββ4.2 Include 3 parts in each test name #new
ββ4.3 Structure tests by the AAA pattern #strategic
ββ4.4 Detect code issues with a linter
ββ4.5 Avoid global test fixtures and seeds, add data per-test #strategic
ββ4.6 Constantly inspect for vulnerable dependencies
ββ4.7 Tag your tests #advanced
ββ4.8 Check your test coverage, it helps to identify wrong test patterns
ββ4.9 Inspect for outdated packages
ββ4.10 Use production-like environment for e2e testing
ββ4.11 Refactor regularly using static analysis tools
ββ4.12 Carefully choose your CI platform (Jenkins vs CircleCI vs Travis vs Rest of the world)
ββ4.13 Test your middlewares in isolation
5. Going To Production Practices (19)
ββ5.1. Monitoring #strategic
ββ5.2. Increase transparency using smart logging #strategic
ββ5.3. Delegate anything possible (e.g. gzip, SSL) to a reverse proxy #strategic
ββ5.4. Lock dependencies
ββ5.5. Guard process uptime using the right tool
ββ5.6. Utilize all CPU cores
ββ5.7. Create a βmaintenance endpointβ
ββ5.8. Discover errors and downtime using APM products #advanced
ββ5.9. Make your code production-ready
ββ5.10. Measure and guard the memory usage #advanced
ββ5.11. Get your frontend assets out of Node
ββ5.12. Be stateless, kill your servers almost every day
ββ5.13. Use tools that automatically detect vulnerabilities
ββ5.14. Assign a transaction id to each log statement #advanced
ββ5.15. Set NODE_ENV=production
ββ5.16. Design automated, atomic and zero-downtime deployments #advanced
ββ5.17. Use an LTS release of Node.js
ββ5.18. Don't route logs within the app
ββ5.19. Install your packages with npm ci #new
6. Security Practices (25)
ββ6.1. Embrace linter security rules
ββ6.2. Limit concurrent requests using a middleware
ββ6.3 Extract secrets from config files or use packages to encrypt them #strategic
ββ6.4. Prevent query injection vulnerabilities with ORM/ODM libraries #strategic
ββ6.5. Collection of generic security best practices
ββ6.6. Adjust the HTTP response headers for enhanced security
ββ6.7. Constantly and automatically inspect for vulnerable dependencies #strategic
ββ6.8. Protect Users' Passwords/Secrets using bcrypt or scrypt #strategic
ββ6.9. Escape HTML, JS and CSS output
ββ6.10. Validate incoming JSON schemas #strategic
ββ6.11. Support blocklisting JWTs
ββ6.12. Prevent brute-force attacks against authorization #advanced
ββ6.13. Run Node.js as non-root user
ββ6.14. Limit payload size using a reverse-proxy or a middleware
ββ6.15. Avoid JavaScript eval statements
ββ6.16. Prevent evil RegEx from overloading your single thread execution
ββ6.17. Avoid module loading using a variable
ββ6.18. Run unsafe code in a sandbox
ββ6.19. Take extra care when working with child processes #advanced
ββ6.20. Hide error details from clients
ββ6.21. Configure 2FA for npm or Yarn #strategic
ββ6.22. Modify session middleware settings
ββ6.23. Avoid DOS attacks by explicitly setting when a process should crash #advanced
ββ6.24. Prevent unsafe redirects
ββ6.25. Avoid publishing secrets to the npm registry
7. Performance Practices (2) (Work In ProgressοΈ βοΈ )
ββ7.1. Don't block the event loop
ββ7.2. Prefer native JS methods over user-land utils like Lodash
8. Docker Practices (15)
ββ8.1 Use multi-stage builds for leaner and more secure Docker images #strategic
ββ8.2. Bootstrap using node command, avoid npm start
ββ8.3. Let the Docker runtime handle replication and uptime #strategic
ββ8.4. Use .dockerignore to prevent leaking secrets
ββ8.5. Clean-up dependencies before production
ββ8.6. Shutdown smartly and gracefully #advanced
ββ8.7. Set memory limits using both Docker and v8 #advanced #strategic
ββ8.8. Plan for efficient caching
ββ8.9. Use explicit image reference, avoid latest tag
ββ8.10. Prefer smaller Docker base images
ββ8.11. Clean-out build-time secrets, avoid secrets in args #strategic #new
ββ8.12. Scan images for multi layers of vulnerabilities #advanced
ββ8.13 Clean NODE_MODULE cache
ββ8.14. Generic Docker practices
ββ8.15. Lint your Dockerfile #new
1. Project Structure Practices
1.1 Structure your solution by components
TL;DR: The worst large applications pitfall is maintaining a huge code base with hundreds of dependencies - such a monolith slows down developers as they try to incorporate new features. Instead, partition your code into components, each gets its folder or a dedicated codebase, and ensure that each unit is kept small and simple. Visit 'Read More' below to see examples of correct project structure
Otherwise: When developers who code new features struggle to realize the impact of their change and fear to break other dependent components - deployments become slower and riskier. It's also considered harder to scale-out when all the business units are not separated
1.2 Layer your components, keep the web layer within its boundaries
TL;DR: Each component should contain 'layers' - a dedicated object for the web, logic, and data access code. This not only draws a clean separation of concerns but also significantly eases mocking and testing the system. Though this is a very common pattern, API developers tend to mix layers by passing the web layer objects (e.g. Express req, res) to business logic and data layers - this makes your application dependent on and accessible only by specific web frameworks
Otherwise: App that mixes web objects with other layers cannot be accessed by testing code, CRON jobs, triggers from message queues, etc
1.3 Wrap common utilities as npm packages
TL;DR: In a large app that constitutes a large codebase, cross-cutting-concern utilities like a logger, encryption and alike, should be wrapped by your code and exposed as private npm packages. This allows sharing them among multiple codebases and projects
Otherwise: You'll have to invent your deployment and the dependency wheel
1.4 Separate Express 'app' and 'server'
TL;DR: Avoid the nasty habit of defining the entire Express app in a single huge file - separate your 'Express' definition to at least two files: the API declaration (app.js) and the networking concerns (WWW). For even better structure, locate your API declaration within components
Otherwise: Your API will be accessible for testing via HTTP calls only (slower and much harder to generate coverage reports). It probably won't be a big pleasure to maintain hundreds of lines of code in a single file
1.5 Use environment aware, secure and hierarchical config
TL;DR: A perfect and flawless configuration setup should ensure (a) keys can be read from file AND from environment variable (b) secrets are kept outside committed code (c) config is hierarchical for easier findability. There are a few packages that can help tick most of those boxes like rc, nconf, config, and convict.
Otherwise: Failing to satisfy any of the config requirements will simply bog down the development or DevOps team. Probably both
2. Error Handling Practices
2.1 Use Async-Await or promises for async error handling
TL;DR: Handling async errors in callback style is probably the fastest way to hell (a.k.a the pyramid of doom). The best gift you can give to your code is using a reputable promise library or async-await instead which enables a much more compact and familiar code syntax like try-catch
Otherwise: Node.js callback style, function(err, response), is a promising way to un-maintainable code due to the mix of error handling with casual code, excessive nesting, and awkward coding patterns
2.2 Use only the built-in Error object
TL;DR: Many throw errors as a string or as some custom type β this complicates the error handling logic and the interoperability between modules. Whether you reject a promise, throw an exception or emit an error β using only the built-in Error object (or an object that extends the built-in Error object) will increase uniformity and prevent loss of information. There is no-throw-literal ESLint rule that strictly checks that (although it have some limitations which can be solved when using TypeScript and setting the @typescript-eslint/no-throw-literal rule)
Otherwise: When invoking some component, being uncertain which type of errors come in return β it makes proper error handling much harder. Even worse, using custom types to describe errors might lead to loss of critical error information like the stack trace!
2.3 Distinguish operational vs programmer errors
TL;DR: Operational errors (e.g. API received an invalid input) refer to known cases where the error impact is fully understood and can be handled thoughtfully. On the other hand, programmer error (e.g. trying to read an undefined variable) refers to unknown code failures that dictate to gracefully restart the application
Otherwise: You may always restart the application when an error appears, but why let ~5000 online users down because of a minor, predicted, operational error? The opposite is also not ideal β keeping the application up when an unknown issue (programmer error) occurred might lead to an unpredicted behavior. Differentiating the two allows acting tactfully and applying a balanced approach based on the given context
2.4 Handle errors centrally, not within a middleware
TL;DR: Error handling logic such as mail to admin and logging should be encapsulated in a dedicated and centralized object that all endpoints (e.g. Express middleware, cron jobs, unit-testing) call when an error comes in
Otherwise: Not handling errors within a single place will lead to code duplication and probably to improperly handled errors
2.5 Document API errors using Swagger or GraphQL
TL;DR: Let your API callers know which errors might come in return so they can handle these thoughtfully without crashing. For RESTful APIs, this is usually done with documentation frameworks like Swagger. If you're using GraphQL, you can utilize your schema and comments as well.
Otherwise: An API client might decide to crash and restart only because it received back an error it couldnβt understand. Note: the caller of your API might be you (very typical in a microservice environment)
2.6 Exit the process gracefully when a stranger comes to town
TL;DR: When an unknown error occurs (a developer error, see best practice 2.3) - there is uncertainty about the application healthiness. Common practice suggests restarting the process carefully using a process management tool like Forever or PM2
Otherwise: When an unfamiliar exception occurs, some object might be in a faulty state (e.g. an event emitter which is used globally and not firing events anymore due to some internal failure) and all future requests might fail or behave crazily
2.7 Use a mature logger to increase error visibility
TL;DR: A set of mature logging tools like Pino or Log4js, will speed-up error discovery and understanding. So forget about console.log
Otherwise: Skimming through console.logs or manually through messy text file without querying tools or a decent log viewer might keep you busy at work until late
2.8 Test error flows using your favorite test framework
TL;DR: Whether professional automated QA or plain manual developer testing β Ensure that your code not only satisfies positive scenarios but also handles and returns the right errors. Testing frameworks like Mocha & Chai can handle this easily (see code examples within the "Gist popup")
Otherwise: Without testing, whether automatically or manually, you canβt rely on your code to return the right errors. Without meaningful errors β thereβs no error handling
2.9 Discover errors and downtime using APM products
TL;DR: Monitoring and performance products (a.k.a APM) proactively gauge your codebase or API so they can automagically highlight errors, crashes, and slow parts that you were missing
Otherwise: You might spend great effort on measuring API performance and downtimes, probably youβll never be aware which are your slowest code parts under real-world scenario and how these affect the UX
2.10 Catch unhandled promise rejections
TL;DR: Any exception thrown within a promise will get swallowed and discarded unless a developer didnβt forget to explicitly handle it. Even if your code is subscribed to process.uncaughtException! Overcome this by registering to the event process.unhandledRejection
Otherwise: Your errors will get swallowed and leave no trace. Nothing to worry about
2.11 Fail fast, validate arguments using a dedicated library
TL;DR: Assert API input to avoid nasty bugs that are much harder to track later. The validation code is usually tedious unless you are using a very cool helper library like ajv and Joi
Otherwise: Consider this β your function expects a numeric argument βDiscountβ which the caller forgets to pass, later on, your code checks if Discount!=0 (amount of allowed discount is greater than zero), then it will allow the user to enjoy a discount. OMG, what a nasty bug. Can you see it?
2.12 Always await promises before returning to avoid a partial stacktrace
TL;DR: Always do return await when returning a promise to benefit full error stacktrace. If a
function returns a promise, that function must be declared as async function and explicitly
await the promise before returning it
Otherwise: The function that returns a promise without awaiting won't appear in the stacktrace. Such missing frames would probably complicate the understanding of the flow that leads to the error, especially if the cause of the abnormal behavior is inside of the missing function
3. Code Style Practices
3.1 Use ESLint
TL;DR: ESLint is the de-facto standard for checking possible code errors and fixing code style, not only to identify nitty-gritty spacing issues but also to detect serious code anti-patterns like developers throwing errors without classification. Though ESLint can automatically fix code styles, other tools like prettier and beautify are more powerful in formatting the fix and work in conjunction with ESLint
Otherwise: Developers will focus on tedious spacing and line-width concerns and time might be wasted overthinking the project's code style
3.2 Node.js specific plugins
TL;DR: On top of ESLint standard rules that cover vanilla JavaScript, add Node.js specific plugins like eslint-plugin-node, eslint-plugin-mocha and eslint-plugin-node-security
Otherwise: Many faulty Node.js code patterns might escape under the radar. For example, developers might require(variableAsPath) files with a variable given as a path which allows attackers to execute any JS script. Node.js linters can detect such patterns and complain early
3.3 Start a Codeblock's Curly Braces on the Same Line
TL;DR: The opening curly braces of a code block should be on the same line as the opening statement
Code Example
// Do
function someFunction() {
// code block
}
// Avoid
function someFunction()
{
// code block
}Otherwise: Deferring from this best practice might lead to unexpected results, as seen in the StackOverflow thread below:
3.4 Separate your statements properly
No matter if you use semicolons or not to separate your statements, knowing the common pitfalls of improper linebreaks or automatic semicolon insertion, will help you to eliminate regular syntax errors.
TL;DR: Use ESLint to gain awareness about separation concerns. Prettier or Standardjs can automatically resolve these issues.
Otherwise: As seen in the previous section, JavaScript's interpreter automatically adds a semicolon at the end of a statement if there isn't one, or considers a statement as not ended where it should, which might lead to some undesired results. You can use assignments and avoid using immediately invoked function expressions to prevent most of the unexpected errors.
Code example
// Do
function doThing() {
// ...
}
doThing()
// Do
const items = [1, 2, 3]
items.forEach(console.log)
// Avoid β throws exception
const m = new Map()
const a = [1,2,3]
[...m.values()].forEach(console.log)
> [...m.values()].forEach(console.log)
> ^^^
> SyntaxError: Unexpected token ...
// Avoid β throws exception
const count = 2 // it tries to run 2(), but 2 is not a function
(function doSomething() {
// do something amazing
}())
// put a semicolon before the immediate invoked function, after the const definition, save the return value of the anonymous function to a variable or avoid IIFEs altogether 3.5 Name your functions
TL;DR: Name all functions, including closures and callbacks. Avoid anonymous functions. This is especially useful when profiling a node app. Naming all functions will allow you to easily understand what you're looking at when checking a memory snapshot
Otherwise: Debugging production issues using a core dump (memory snapshot) might become challenging as you notice significant memory consumption from anonymous functions
3.6 Use naming conventions for variables, constants, functions and classes
TL;DR: Use lowerCamelCase when naming constants, variables and functions, UpperCamelCase (capital first letter as well) when naming classes and UPPER_SNAKE_CASE when naming global or static variables. This will help you to easily distinguish between plain variables, functions, classes that require instantiation and variables declared at global module scope. Use descriptive names, but try to keep them short
Otherwise: JavaScript is the only language in the world that allows invoking a constructor ("Class") directly without instantiating it first. Consequently, Classes and function-constructors are differentiated by starting with UpperCamelCase
3.6 Code Example
// for global variables names we use the const/let keyword and UPPER_SNAKE_CASE
let MUTABLE_GLOBAL = "mutable value"
const GLOBAL_CONSTANT = "immutable value";
const CONFIG = {
key: "value",
};
// examples of UPPER_SNAKE_CASE convention in nodejs/javascript ecosystem
// in javascript Math.PI module
const PI = 3.141592653589793;
// https://github.com/nodejs/node/blob/b9f36062d7b5c5039498e98d2f2c180dca2a7065/lib/internal/http2/core.js#L303
// in nodejs http2 module
const HTTP_STATUS_OK = 200;
const HTTP_STATUS_CREATED = 201;
// for class name we use UpperCamelCase
class SomeClassExample {
// for static class properties we use UPPER_SNAKE_CASE
static STATIC_PROPERTY = "value";
}
// for functions names we use lowerCamelCase
function doSomething() {
// for scoped variable names we use the const/let keyword and lowerCamelCase
const someConstExample = "immutable value";
let someMutableExample = "mutable value";
} 3.7 Prefer const over let. Ditch the var
TL;DR: Using const means that once a variable is assigned, it cannot be reassigned. Preferring const will help you to not be tempted to use the same variable for different uses, and make your code clearer. If a variable needs to be reassigned, in a for loop, for example, use let to declare it. Another important aspect of let is that a variable declared using it is only available in the block scope in which it was defined. var is function scoped, not block-scoped, and shouldn't be used in ES6 now that you have const and let at your disposal
Otherwise: Debugging becomes way more cumbersome when following a variable that frequently changes
3.8 Require modules first, not inside functions
TL;DR: Require modules at the beginning of each file, before and outside of any functions. This simple best practice will not only help you easily and quickly tell the dependencies of a file right at the top but also avoids a couple of potential problems
Otherwise: Requires are run synchronously by Node.js. If they are called from within a function, it may block other requests from being handled at a more critical time. Also, if a required module or any of its dependencies throw an error and crash the server, it is best to find out about it as soon as possible, which might not be the case if that module is required from within a function
3.9 Require modules by folders, as opposed to the files directly
TL;DR: When developing a module/library in a folder, place an index.js file that exposes the module's internals so every consumer will pass through it. This serves as an 'interface' to your module and eases future changes without breaking the contract
Otherwise: Changing the internal structure of files or the signature may break the interface with clients
3.9 Code example
// Do
module.exports.SMSProvider = require("./SMSProvider");
module.exports.SMSNumberResolver = require("./SMSNumberResolver");
// Avoid
module.exports.SMSProvider = require("./SMSProvider/SMSProvider.js");
module.exports.SMSNumberResolver = require("./SMSNumberResolver/SMSNumberResolver.js"); 3.10 Use the === operator
TL;DR: Prefer the strict equality operator === over the weaker abstract equality operator ==. == will compare two variables after converting them to a common type. There is no type conversion in ===, and both variables must be of the same type to be equal
Otherwise: Unequal variables might return true when compared with the == operator
3.10 Code example
"" == "0"; // false
0 == ""; // true
0 == "0"; // true
false == "false"; // false
false == "0"; // true
false == undefined; // false
false == null; // false
null == undefined; // true
" \t\r\n " == 0; // trueAll statements above will return false if used with ===
3.11 Use Async Await, avoid callbacks
TL;DR: Node 8 LTS now has full support for Async-await. This is a new way of dealing with asynchronous code which supersedes callbacks and promises. Async-await is non-blocking, and it makes asynchronous code look synchronous. The best gift you can give to your code is using async-await which provides a much more compact and familiar code syntax like try-catch
Otherwise: Handling async errors in callback style are probably the fastest way to hell - this style forces to check errors all over, deal with awkward code nesting, and makes it difficult to reason about the code flow
3.12 Use arrow function expressions (=>)
TL;DR: Though it's recommended to use async-await and avoid function parameters when dealing with older APIs that accept promises or callbacks - arrow functions make the code structure more compact and keep the lexical context of the root function (i.e. this)
Otherwise: Longer code (in ES5 functions) is more prone to bugs and cumbersome to read
4. Testing And Overall Quality Practices
4.1 At the very least, write API (component) testing
TL;DR: Most projects just don't have any automated testing due to short timetables or often the 'testing project' ran out of control and was abandoned. For that reason, prioritize and start with API testing which is the easiest way to write and provides more coverage than unit testing (you may even craft API tests without code using tools like Postman). Afterwards, should you have more resources and time, continue with advanced test types like unit testing, DB testing, performance testing, etc
Otherwise: You may spend long days on writing unit tests to find out that you got only 20% system coverage
4.2 Include 3 parts in each test name
TL;DR: Make the test speak at the requirements level so it's self-explanatory also to QA engineers and developers who are not familiar with the code internals. State in the test name what is being tested (unit under test), under what circumstances, and what is the expected result
Otherwise: A deployment just failed, a test named βAdd productβ failed. Does this tell you what exactly is malfunctioning?
4.3 Structure tests by the AAA pattern
TL;DR: Structure your tests with 3 well-separated sections: Arrange, Act & Assert (AAA). The first part includes the test setup, then the execution of the unit under test, and finally the assertion phase. Following this structure guarantees that the reader spends no brain CPU on understanding the test plan
Otherwise: Not only you spend long daily hours on understanding the main code, but now also what should have been the simple part of the day (testing) stretches your brain
4.4 Detect code issues with a linter
TL;DR: Use a code linter to check the basic quality and detect anti-patterns early. Run it before any test and add it as a pre-commit git-hook to minimize the time needed to review and correct any issue. Also check Section 3 on Code Style Practices
Otherwise: You may let pass some anti-pattern and possible vulnerable code to your production environment.
4.5 Avoid global test fixtures and seeds, add data per-test
TL;DR: To prevent test coupling and easily reason about the test flow, each test should add and act on its own set of DB rows. Whenever a test needs to pull or assume the existence of some DB data - it must explicitly add that data and avoid mutating any other records
Otherwise: Consider a scenario where deployment is aborted due to failing tests, team is now going to spend precious investigation time that ends in a sad conclusion: the system works well, the tests however interfere with each other and break the build
4.6 Constantly inspect for vulnerable dependencies
TL;DR: Even the most reputable dependencies such as Express have known vulnerabilities. This can get easily tamed using community and commercial tools such as
Otherwise: Keeping your code clean from vulnerabilities without dedicated tools will require to constantly follow online publications about new threats. Quite tedious
4.7 Tag your tests
TL;DR: Different tests must run on different scenarios: quick smoke, IO-less, tests should run when a developer saves or commits a file, full end-to-end tests usually run when a new pull request is submitted, etc. This can be achieved by tagging tests with keywords like #cold #api #sanity so you can grep with your testing harness and invoke the desired subset. For example, this is how you would invoke only the sanity test group with Mocha: mocha --grep 'sanity'
Otherwise: Running all the tests, including tests that perform dozens of DB queries, any time a developer makes a small change can be extremely slow and keeps developers away from running tests
4.8 Check your test coverage, it helps to identify wrong test patterns
TL;DR: Code coverage tools like Istanbul/NYC are great for 3 reasons: it comes for free (no effort is required to benefit this reports), it helps to identify a decrease in testing coverage, and last but not least it highlights testing mismatches: by looking at colored code coverage reports you may notice, for example, code areas that are never tested like catch clauses (meaning that tests only invoke the happy paths and not how the app behaves on errors). Set it to fail builds if the coverage falls under a certain threshold
Otherwise: There won't be any automated metric telling you when a large portion of your code is not covered by testing
4.9 Inspect for outdated packages
TL;DR: Use your preferred tool (e.g. npm outdated or npm-check-updates) to detect installed outdated packages, inject this check into your CI pipeline and even make a build fail in a severe scenario. For example, a severe scenario might be when an installed package is 5 patch commits behind (e.g. local version is 1.3.1 and repository version is 1.3.8) or it is tagged as deprecated by its author - kill the build and prevent deploying this version
Otherwise: Your production will run packages that have been explicitly tagged by their author as risky
4.10 Use production-like environment for e2e testing
TL;DR: End to end (e2e) testing which includes live data used to be the weakest link of the CI process as it depends on multiple heavy services like DB. Use an environment which is as close to your real production environment as possible like a-continue (Missed -continue here, needs content. Judging by the Otherwise clause, this should mention docker-compose)
Otherwise: Without docker-compose, teams must maintain a testing DB for each testing environment including developers' machines, keep all those DBs in sync so test results won't vary across environments
4.11 Refactor regularly using static analysis tools
TL;DR: Using static analysis tools helps by giving objective ways to improve code quality and keeps your code maintainable. You can add static analysis tools to your CI build to fail when it finds code smells. Its main selling points over plain linting are the ability to inspect quality in the context of multiple files (e.g. detect duplications), perform advanced analysis (e.g. code complexity), and follow the history and progress of code issues. Two examples of tools you can use are Sonarqube (2,600+ stars) and Code Climate (1,500+ stars).
Otherwise: With poor code quality, bugs and performance will always be an issue that no shiny new library or state of the art features can fix
4.12 Carefully choose your CI platform (Jenkins vs CircleCI vs Travis vs Rest of the world)
TL;DR: Your continuous integration platform (CICD) will host all the quality tools (e.g. test, lint) so it should come with a vibrant ecosystem of plugins. Jenkins used to be the default for many projects as it has the biggest community along with a very powerful platform at the price of a complex setup that demands a steep learning curve. Nowadays, it has become much easier to set up a CI solution using SaaS tools like CircleCI and others. These tools allow crafting a flexible CI pipeline without the burden of managing the whole infrastructure. Eventually, it's a trade-off between robustness and speed - choose your side carefully
Otherwise: Choosing some niche vendor might get you blocked once you need some advanced customization. On the other hand, going with Jenkins might burn precious time on infrastructure setup
4.13 Test your middlewares in isolation
TL;DR: When a middleware holds some immense logic that spans many requests, it is worth testing it in isolation without waking up the entire web framework. This can be easily achieved by stubbing and spying on the {req, res, next} objects
Otherwise: A bug in Express middleware === a bug in all or most requests
5. Going To Production Practices
5.1. Monitoring
TL;DR: Monitoring is a game of finding out issues before customers do β obviously this should be assigned unprecedented importance. The market is overwhelmed with offers thus consider starting with defining the basic metrics you must follow (my suggestions inside), then go over additional fancy features and choose the solution that ticks all boxes. Click βThe Gistβ below for an overview of the solutions
Otherwise: Failure === disappointed customers. Simple
5.2. Increase transparency using smart logging
TL;DR: Logs can be a dumb warehouse of debug statements or the enabler of a beautiful dashboard that tells the story of your app. Plan your logging platform from day 1: how logs are collected, stored and analyzed to ensure that the desired information (e.g. error rate, following an entire transaction through services and servers, etc) can really be extracted
Otherwise: You end up with a black box that is hard to reason about, then you start re-writing all logging statements to add additional information
5.3. Delegate anything possible (e.g. gzip, SSL) to a reverse proxy
TL;DR: Node is awfully bad at doing CPU intensive tasks like gzipping, SSL termination, etc. You should use βrealβ middleware services like nginx, HAproxy or cloud vendor services instead
Otherwise: Your poor single thread will stay busy doing infrastructural tasks instead of dealing with your application core and performance will degrade accordingly
5.4. Lock dependencies
TL;DR: Your code must be identical across all environments, but amazingly npm lets dependencies drift across environments by default β when you install packages at various environments it tries to fetch packagesβ latest patch version. Overcome this by using npm config files, .npmrc, that tell each environment to save the exact (not the latest) version of each package. Alternatively, for finer grained control use npm shrinkwrap. *Update: as of NPM5, dependencies are locked by default. The new package manager in town, Yarn, also got us covered by default
Otherwise: QA will thoroughly test the code and approve a version that will behave differently in production. Even worse, different servers in the same production cluster might run different code
5.5. Guard process uptime using the right tool
TL;DR: The process must go on and get restarted upon failures. For simple scenarios, process management tools like PM2 might be enough but in today's βdockerizedβ world, cluster management tools should be considered as well
Otherwise: Running dozens of instances without a clear strategy and too many tools together (cluster management, docker, PM2) might lead to DevOps chaos
5.6. Utilize all CPU cores
TL;DR: At its basic form, a Node app runs on a single CPU core while all others are left idling. Itβs your duty to replicate the Node process and utilize all CPUs β For small-medium apps you may use Node Cluster or PM2. For a larger app consider replicating the process using some Docker cluster (e.g. K8S, ECS) or deployment scripts that are based on Linux init system (e.g. systemd)
Otherwise: Your app will likely utilize only 25% of its available resources(!) or even less. Note that a typical server has 4 CPU cores or more, naive deployment of Node.js utilizes only 1 (even using PaaS services like AWS beanstalk!)
5.7. Create a βmaintenance endpointβ
TL;DR: Expose a set of system-related information, like memory usage and REPL, etc in a secured API. Although itβs highly recommended to rely on standard and battle-tested tools, some valuable information and operations are easier done using code
Otherwise: Youβll find that youβre performing many βdiagnostic deploysβ β shipping code to production only to extract some information for diagnostic purposes
5.8. Discover errors and downtime using APM products
TL;DR: Application monitoring and performance products (a.k.a. APM) proactively gauge codebase and API so they can auto-magically go beyond traditional monitoring and measure the overall user-experience across services and tiers. For example, some APM products can highlight a transaction that loads too slow on the end-user's side while suggesting the root cause
Otherwise: You might spend great effort on measuring API performance and downtimes, probably youβll never be aware which is your slowest code parts under real-world scenario and how these affect the UX
5.9. Make your code production-ready
TL;DR: Code with the end in mind, plan for production from day 1. This sounds a bit vague so Iβve compiled a few development tips that are closely related to production maintenance (click Gist below)
Otherwise: A world champion IT/DevOps guy wonβt save a system that is badly written
5.10. Measure and guard the memory usage
TL;DR: Node.js has controversial relationships with memory: the v8 engine has soft limits on memory usage (1.4GB) and there are known paths to leak memory in Nodeβs code β thus watching Nodeβs process memory is a must. In small apps, you may gauge memory periodically using shell commands but in medium-large apps consider baking your memory watch into a robust monitoring system
Otherwise: Your process memory might leak a hundred megabytes a day like how it happened at Walmart
5.11. Get your frontend assets out of Node
TL;DR: Serve frontend content using dedicated middleware (nginx, S3, CDN) because Node performance really gets hurt when dealing with many static files due to its single-threaded model
Otherwise: Your single Node thread will be busy streaming hundreds of html/images/angular/react files instead of allocating all its resources for the task it was born for β serving dynamic content
5.12. Be stateless, kill your servers almost every day
TL;DR: Store any type of data (e.g. user sessions, cache, uploaded files) within external data stores. Consider βkillingβ your servers periodically or use βserverlessβ platform (e.g. AWS Lambda) that explicitly enforces a stateless behavior
Otherwise: Failure at a given server will result in application downtime instead of just killing a faulty machine. Moreover, scaling-out elasticity will get more challenging due to the reliance on a specific server
5.13. Use tools that automatically detect vulnerabilities
TL;DR: Even the most reputable dependencies such as Express have known vulnerabilities (from time to time) that can put a system at risk. This can be easily tamed using community and commercial tools that constantly check for vulnerabilities and warn (locally or at GitHub), some can even patch them immediately
Otherwise: Keeping your code clean from vulnerabilities without dedicated tools will require you to constantly follow online publications about new threats. Quite tedious
5.14. Assign a transaction id to each log statement
Also known as correlation id / transit id / tracing id / request id / request context / etc.
TL;DR: Assign the same identifier, transaction-id: {some value}, to each log entry within a single request. Then when inspecting errors in logs, easily conclude what happened before and after. Until version 14 of Node, this was not easy to achieve due to Node's async nature, but since AsyncLocalStorage came to town, this became possible and easy than ever. see code examples inside
Otherwise: Looking at a production error log without the context β what happened before β makes it much harder and slower to reason about the issue
5.15. Set NODE_ENV=production
TL;DR: Set the environment variable NODE_ENV to βproductionβ or βdevelopmentβ to flag whether production optimizations should get activated β many npm packages determine the current environment and optimize their code for production
Otherwise: Omitting this simple property might greatly degrade performance. For example, when using Express for server-side rendering omitting NODE_ENV makes it slower by a factor of three!
5.16. Design automated, atomic and zero-downtime deployments
TL;DR: Research shows that teams who perform many deployments lower the probability of severe production issues. Fast and automated deployments that donβt require risky manual steps and service downtime significantly improve the deployment process. You should probably achieve this using Docker combined with CI tools as they became the industry standard for streamlined deployment
Otherwise: Long deployments -> production downtime & human-related error -> team unconfident in making deployment -> fewer deployments and features
5.17. Use an LTS release of Node.js
TL;DR: Ensure you are using an LTS version of Node.js to receive critical bug fixes, security updates and performance improvements
Otherwise: Newly discovered bugs or vulnerabilities could be used to exploit an application running in production, and your application may become unsupported by various modules and harder to maintain
5.18. Don't route logs within the app
TL;DR: Log destinations should not be hard-coded by developers within the application code, but instead should be defined by the execution environment the application runs in. Developers should write logs to stdout using a logger utility and then let the execution environment (container, server, etc.) pipe the stdout stream to the appropriate destination (i.e. Splunk, Graylog, ElasticSearch, etc.).
Otherwise: Application handling log routing === hard to scale, loss of logs, poor separation of concerns
5.19. Install your packages with npm ci
TL;DR: You have to be sure that production code uses the exact version of the packages you have tested it with. Run npm ci to strictly do a clean install of your dependencies matching package.json and package-lock.json. Using this command is recommended in automated environments such as continuous integration pipelines.
Otherwise: QA will thoroughly test the code and approve a version that will behave differently in production. Even worse, different servers in the same production cluster might run different code.
6. Security Best Practices
6.1. Embrace linter security rules
TL;DR: Make use of security-related linter plugins such as eslint-plugin-security to catch security vulnerabilities and issues as early as possible, preferably while they're being coded. This can help catching security weaknesses like using eval, invoking a child process or importing a module with a string literal (e.g. user input). Click 'Read more' below to see code examples that will get caught by a security linter
Otherwise: What could have been a straightforward security weakness during development becomes a major issue in production. Also, the project may not follow consistent code security practices, leading to vulnerabilities being introduced, or sensitive secrets committed into remote repositories
6.2. Limit concurrent requests using a middleware
TL;DR: DOS attacks are very popular and relatively easy to conduct. Implement rate limiting using an external service such as cloud load balancers, cloud firewalls, nginx, rate-limiter-flexible package, or (for smaller and less critical apps) a rate-limiting middleware (e.g. express-rate-limit)
Otherwise: An application could be subject to an attack resulting in a denial of service where real users receive a degraded or unavailable service.
6.3 Extract secrets from config files or use packages to encrypt them
TL;DR: Never store plain-text secrets in configuration files or source code. Instead, make use of secret-management systems like Vault products, Kubernetes/Docker Secrets, or using environment variables. As a last resort, secrets stored in source control must be encrypted and managed (rolling keys, expiring, auditing, etc). Make use of pre-commit/push hooks to prevent committing secrets accidentally
Otherwise: Source control, even for private repositories, can mistakenly be made public, at which point all secrets are exposed. Access to source control for an external party will inadvertently provide access to related systems (databases, apis, services, etc).
6.4. Prevent query injection vulnerabilities with ORM/ODM libraries
TL;DR: To prevent SQL/NoSQL injection and other malicious attacks, always make use of an ORM/ODM or a database library that escapes data or supports named or indexed parameterized queries, and takes care of validating user input for expected types. Never just use JavaScript template strings or string concatenation to inject values into queries as this opens your application to a wide spectrum of vulnerabilities. All the reputable Node.js data access libraries (e.g. Sequelize, Knex, mongoose) have built-in protection against injection attacks.
Otherwise: Unvalidated or unsanitized user input could lead to operator injection when working with MongoDB for NoSQL, and not using a proper sanitization system or ORM will easily allow SQL injection attacks, creating a giant vulnerability.
6.5. Collection of generic security best practices
TL;DR: This is a collection of security advice that is not related directly to Node.js - the Node implementation is not much different than any other language. Click read more to skim through.
6.6. Adjust the HTTP response headers for enhanced security
TL;DR: Your application should be using secure headers to prevent attackers from using common attacks like cross-site scripting (XSS), clickjacking and other malicious attacks. These can be configured easily using modules like helmet.
Otherwise: Attackers could perform direct attacks on your application's users, leading to huge security vulnerabilities
6.7. Constantly and automatically inspect for vulnerable dependencies
TL;DR: With the npm ecosystem it is common to have many dependencies for a project. Dependencies should always be kept in check as new vulnerabilities are found. Use tools like npm audit or snyk to track, monitor and patch vulnerable dependencies. Integrate these tools with your CI setup so you catch a vulnerable dependency before it makes it to production.
Otherwise: An attacker could detect your web framework and attack all its known vulnerabilities.
6.8. Protect Users' Passwords/Secrets using bcrypt or scrypt
TL;DR: Passwords or secrets (e.g. API keys) should be stored using a secure hash + salt function like bcrypt,scrypt, or worst case pbkdf2.
Otherwise: Passwords and secrets that are stored without using a secure function are vulnerable to brute forcing and dictionary attacks that will lead to their disclosure eventually.
6.9. Escape HTML, JS and CSS output
TL;DR: Untrusted data that is sent down to the browser might get executed instead of just being displayed, this is commonly referred as a cross-site-scripting (XSS) attack. Mitigate this by using dedicated libraries that explicitly mark the data as pure content that should never get executed (i.e. encoding, escaping)
Otherwise: An attacker might store malicious JavaScript code in your DB which will then be sent as-is to the poor clients
6.10. Validate incoming JSON schemas
TL;DR: Validate the incoming requests' body payload and ensure it meets expectations, fail fast if it doesn't. To avoid tedious validation coding within each route you may use lightweight JSON-based validation schemas such as jsonschema or joi
Otherwise: Your generosity and permissive approach greatly increases the attack surface and encourages the attacker to try out many inputs until they find some combination to crash the application
6.11. Support blocklisting JWTs
TL;DR: When using JSON Web Tokens (for example, with Passport.js), by default there's no mechanism to revoke access from issued tokens. Once you discover some malicious user activity, there's no way to stop them from accessing the system as long as they hold a valid token. Mitigate this by implementing a blocklist of untrusted tokens that are validated on each request.
Otherwise: Expired, or misplaced tokens could be used maliciously by a third party to access an application and impersonate the owner of the token.
6.12. Prevent brute-force attacks against authorization
TL;DR: A simple and powerful technique is to limit authorization attempts using two metrics:
- The first is number of consecutive failed attempts by the same user unique ID/name and IP address.
- The second is number of failed attempts from an IP address over some long period of time. For example, block an IP address if it makes 100 failed attempts in one day.
Otherwise: An attacker can issue unlimited automated password attempts to gain access to privileged accounts on an application
6.13. Run Node.js as non-root user
TL;DR: There is a common scenario where Node.js runs as a root user with unlimited permissions. For example, this is the default behaviour in Docker containers. It's recommended to create a non-root user and either bake it into the Docker image (examples given below) or run the process on this user's behalf by invoking the container with the flag "-u username"
Otherwise: An attacker who manages to run a script on the server gets unlimited power over the local machine (e.g. change iptable and re-route traffic to their server)
6.14. Limit payload size using a reverse-proxy or a middleware
TL;DR: The bigger the body payload is, the harder your single thread works in processing it. This is an opportunity for attackers to bring servers to their knees without tremendous amount of requests (DOS/DDOS attacks). Mitigate this limiting the body size of incoming requests on the edge (e.g. firewall, ELB) or by configuring express body parser to accept only small-size payloads
Otherwise: Your application will have to deal with large requests, unable to process the other important work it has to accomplish, leading to performance implications and vulnerability towards DOS attacks
6.15. Avoid JavaScript eval statements
TL;DR: eval is evil as it allows executing custom JavaScript code during run time. This is not just a performance concern but also an important security concern due to malicious JavaScript code that may be sourced from user input. Another language feature that should be avoided is new Function constructor. setTimeout and setInterval should never be passed dynamic JavaScript code either.
Otherwise: Malicious JavaScript code finds a way into text passed into eval or other real-time evaluating JavaScript language functions, and will gain complete access to JavaScript permissions on the page. This vulnerability is often manifested as an XSS attack.
6.16. Prevent evil RegEx from overloading your single thread execution
TL;DR: Regular Expressions, while being handy, pose a real threat to JavaScript applications at large, and the Node.js platform in particular. A user input for text to match might require an outstanding amount of CPU cycles to process. RegEx processing might be inefficient to an extent that a single request that validates 10 words can block the entire event loop for 6 seconds and set the CPU on
Otherwise: Poorly written regexes could be susceptible to Regular Expression DoS attacks that will block the event loop completely. For example, the popular moment package was found vulnerable with malicious RegEx usage in November of 2017
6.17. Avoid module loading using a variable
TL;DR: Avoid requiring/importing another file with a path that was given as parameter due to the concern that it could have originated from user input. This rule can be extended for accessing files in general (i.e. fs.readFile()) or other sensitive resource access with dynamic variables originating from user input. Eslint-plugin-security linter can catch such patterns and warn early enough
Otherwise: Malicious user input could find its way to a parameter that is used to require tampered files, for example, a previously uploaded file on the file system, or access already existing system files.
6.18. Run unsafe code in a sandbox
TL;DR: When tasked to run external code that is given at run-time (e.g. plugin), use any sort of 'sandbox' execution environment that isolates and guards the main code against the plugin. This can be achieved using a dedicated process (e.g. cluster.fork()), serverless environment or dedicated npm packages that act as a sandbox
Otherwise: A plugin can attack through an endless variety of options like infinite loops, memory overloading, and access to sensitive process environment variables
6.19. Take extra care when working with child processes
TL;DR: Avoid using child processes when possible and validate and sanitize input to mitigate shell injection attacks if you still have to. Prefer using child_process.execFile which by definition will only execute a single command with a set of attributes and will not allow shell parameter expansion.
Otherwise: Naive use of child processes could result in remote command execution or shell injection attacks due to malicious user input passed to an unsanitized system command.
6.20. Hide error details from clients
TL;DR: An integrated express error handler hides the error details by default. However, great are the chances that you implement your own error handling logic with custom Error objects (considered by many as a best practice). If you do so, ensure not to return the entire Error object to the client, which might contain some sensitive application details
Otherwise: Sensitive application details such as server file paths, third party modules in use, and other internal workflows of the application which could be exploited by an attacker, could be leaked from information found in a stack trace
6.21. Configure 2FA for npm or Yarn
TL;DR: Any step in the development chain should be protected with MFA (multi-factor authentication), npm/Yarn are a sweet opportunity for attackers who can get their hands on some developer's password. Using developer credentials, attackers can inject malicious code into libraries that are widely installed across projects and services. Maybe even across the web if published in public. Enabling 2-factor-authentication in npm leaves almost zero chances for attackers to alter your package code.
Otherwise: Have you heard about the eslint developer whose password was hijacked?
6.22. Modify session middleware settings
TL;DR: Each web framework and technology has its known weaknessesβ-βtelling an attacker which web framework we use is a great help for them. Using the default settings for session middlewares can expose your app to module- and framework-specific hijacking attacks in a similar way to the X-Powered-By header. Try hiding anything that identifies and reveals your tech stack (E.g. Node.js, express)
Otherwise: Cookies could be sent over insecure connections, and an attacker might use session identification to identify the underlying framework of the web application, as well as module-specific vulnerabilities
6.23. Avoid DOS attacks by explicitly setting when a process should crash
TL;DR: The Node process will crash when errors are not handled. Many best practices even recommend to exit even though an error was caught and got handled. Express, for example, will crash on any asynchronous errorβ-βunless you wrap routes with a catch clause. This opens a very sweet attack spot for attackers who recognize what input makes the process crash and repeatedly send the same request. There's no instant remedy for this but a few techniques can mitigate the pain: Alert with critical severity anytime a process crashes due to an unhandled error, validate the input and avoid crashing the process due to invalid user input, wrap all routes with a catch and consider not to crash when an error originated within a request (as opposed to what happens globally)
Otherwise: This is just an educated guess: given many Node.js applications, if we try passing an empty JSON body to all POST requestsβ-βa handful of applications will crash. At that point, we can just repeat sending the same request to take down the applications with ease
6.24. Prevent unsafe redirects
TL;DR: Redirects that do not validate user input can enable attackers to launch phishing scams, steal user credentials, and perform other malicious actions.
Otherwise: If an attacker discovers that you are not validating external, user-supplied input, they may exploit this vulnerability by posting specially-crafted links on forums, social media, and other public places to get users to click it.
6.25. Avoid publishing secrets to the npm registry
TL;DR: Precautions should be taken to avoid the risk of accidentally publishing secrets to public npm registries. An .npmignore file can be used to ignore specific files or folders, or the files array in package.json can act as an allow list.
Otherwise: Your project's API keys, passwords or other secrets are open to be abused by anyone who comes across them, which may result in financial loss, impersonation, and other risks.
7. Draft: Performance Best Practices
Our contributors are working on this section. Would you like to join?
7.1. Don't block the event loop
TL;DR: Avoid CPU intensive tasks as they will block the mostly single-threaded Event Loop and offload those to a dedicated thread, process or even a different technology based on the context.
Otherwise: As the Event Loop is blocked, Node.js will be unable to handle other request thus causing delays for concurrent users. 3000 users are waiting for a response, the content is ready to be served, but one single request blocks the server from dispatching the results back
7.2. Prefer native JS methods over user-land utils like Lodash
TL;DR: It's often more penalising to use utility libraries like lodash and underscore over native methods as it leads to unneeded dependencies and slower performance.
Bear in mind that with the introduction of the new V8 engine alongside the new ES standards, native methods were improved in such a way that it's now about 50% more performant than utility libraries.
Otherwise: You'll have to maintain less performant projects where you could have simply used what was already available or dealt with a few more lines in exchange of a few more files.
8. Docker Best Practices
8.1 Use multi-stage builds for leaner and more secure Docker images
TL;DR: Use multi-stage build to copy only necessary production artifacts. A lot of build-time dependencies and files are not needed for running your application. With multi-stage builds these resources can be used during build while the runtime environment contains only what's necessary. Multi-stage builds are an easy way to get rid of overweight and security threats.
Otherwise: Larger images will take longer to build and ship, build-only tools might contain vulnerabilities and secrets only meant for the build phase might be leaked.
Example Dockerfile for multi-stage builds
FROM node:14.4.0 AS build
COPY . .
RUN npm ci && npm run build
FROM node:slim-14.4.0
USER node
EXPOSE 8080
COPY --from=build /home/node/app/dist /home/node/app/package.json /home/node/app/package-lock.json ./
RUN npm ci --production
CMD [ "node", "dist/app.js" ] 8.2. Bootstrap using node command, avoid npm start
TL;DR: use CMD ['node','server.js'] to start your app, avoid using npm scripts which don't pass OS signals to the code. This prevents problems with child-processes, signal handling, graceful shutdown and having zombie processes.
Otherwise: When no signals are passed, your code will never be notified about shutdowns. Without this, it will lose its chance to close properly possibly losing current requests and/or data.
Read More: Bootstrap container using node command, avoid npm start
8.3. Let the Docker runtime handle replication and uptime
TL;DR: When using a Docker run time orchestrator (e.g., Kubernetes), invoke the Node.js process directly without intermediate process managers or custom code that replicate the process (e.g. PM2, Cluster module). The runtime platform has the highest amount of data and visibility for making placement decision - It knows best how many processes are needed, how to spread them and what to do in case of crashes
Otherwise: Container keeps crashing due to lack of resources will get restarted indefinitely by the process manager. Should Kubernetes be aware of that, it could relocate it to a different roomy instance
8.4. Use .dockerignore to prevent leaking secrets
TL;DR: Include a .dockerignore file that filters out common secret files and development artifacts. By doing so, you might prevent secrets from leaking into the image. As a bonus the build time will significantly decrease. Also, ensure not to copy all files recursively rather explicitly choose what should be copied to Docker
Otherwise: Common personal secret files like .env, .aws and .npmrc will be shared with anybody with access to the image (e.g. Docker repository)
8.5. Clean-up dependencies before production
TL;DR: Although Dev-Dependencies are sometimes needed during the build and test life-cycle, eventually the image that is shipped to production should be minimal and clean from development dependencies. Doing so guarantees that only necessary code is shipped and the amount of potential attacks (i.e. attack surface) is minimized. When using multi-stage build (see dedicated bullet) this can be achieved by installing all dependencies first and finally running npm ci --production
Otherwise: Many of the infamous npm security breaches were found within development packages (e.g. eslint-scope)
8.6. Shutdown smartly and gracefully
TL;DR: Handle the process SIGTERM event and clean-up all existing connection and resources. This should be done while responding to ongoing requests. In Dockerized runtimes shutting down containers is not a rare event, rather a frequent occurrence that happen as part of routine work. Achieving this demands some thoughtful code to orchestrate several moving parts: The load balancer, keep-alive connections, the HTTP server and other resources
Otherwise: Dying immediately means not responding to thousands of disappointed users
8.7. Set memory limits using both Docker and v8
TL;DR: Always configure a memory limit using both Docker and the JavaScript runtime flags. The Docker limit is needed to make thoughtful container placement decision, the --v8's flag max-old-space is needed to kick off the GC on time and prevent under utilization of memory. Practically, set the v8's old space memory to be a just bit less than the container limit
Otherwise: The docker definition is needed to perform thoughtful scaling decision and prevent starving other citizens. Without also defining the v8's limits, it will under utilize the container resources - Without explicit instructions it crashes when utilizing ~50-60% of its host resources
8.8. Plan for efficient caching
TL;DR: Rebuilding a whole docker image from cache can be nearly instantaneous if done correctly. The less updated instructions should be at the top of your Dockerfile and the ones constantly changing (like app code) should be at the bottom.
Otherwise: Docker build will be very long and consume lot of resources even when making tiny changes
8.9. Use explicit image reference, avoid latest tag
TL;DR: Specify an explicit image digest or versioned label, never refer to latest. Developers are often led to believe that specifying the latest tag will provide them with the most recent image in the repository however this is not the case. Using a digest guarantees that every instance of the service is running exactly the same code.
In addition, referring to an image tag means that the base image is subject to change, as image tags cannot be relied upon for a deterministic install. Instead, if a deterministic install is expected, a SHA256 digest can be used to reference an exact image.
Otherwise: A new version of a base image could be deployed into production with breaking changes, causing unintended application behaviour.
8.10. Prefer smaller Docker base images
TL;DR: Large images lead to higher exposure to vulnerabilities and increased resource consumption. Using leaner Docker images, such as Slim and Alpine Linux variants, mitigates this issue.
Otherwise: Building, pushing, and pulling images will take longer, unknown attack vectors can be used by malicious actors and more resources are consumed.
8.11. Clean-out build-time secrets, avoid secrets in args
TL;DR: Avoid secrets leaking from the Docker build environment. A Docker image is typically shared in multiple environment like CI and a registry that are not as sanitized as production. A typical example is an npm token which is usually passed to a dockerfile as argument. This token stays within the image long after it is needed and allows the attacker indefinite access to a private npm registry. This can be avoided by coping a secret file like .npmrc and then removing it using multi-stage build (beware, build history should be deleted as well) or by using Docker build-kit secret feature which leaves zero traces
Otherwise: Everyone with access to the CI and docker registry will also get access to some precious organization secrets as a bonus
8.12. Scan images for multi layers of vulnerabilities
TL;DR: Besides checking code dependencies vulnerabilities also scan the final image that is shipped to production. Docker image scanners check the code dependencies but also the OS binaries. This E2E security scan covers more ground and verifies that no bad guy injected bad things during the build. Consequently, it is recommended running this as the last step before deployment. There are a handful of free and commercial scanners that also provide CI/CD plugins
Otherwise: Your code might be entirely free from vulnerabilities. However it might still get hacked due to vulnerable version of OS-level binaries (e.g. OpenSSL, TarBall) that are commonly being used by applications
8.13 Clean NODE_MODULE cache
TL;DR: After installing dependencies in a container remove the local cache. It doesn't make any sense to duplicate the dependencies for faster future installs since there won't be any further installs - A Docker image is immutable. Using a single line of code tens of MB (typically 10-50% of the image size) are shaved off
Otherwise: The image that will get shipped to production will weigh 30% more due to files that will never get used
8.14. Generic Docker practices
TL;DR: This is a collection of Docker advice that is not related directly to Node.js - the Node implementation is not much different than any other language. Click read more to skim through.
8.15. Lint your Dockerfile
TL;DR: Linting your Dockerfile is an important step to identify issues in your Dockerfile which differ from best practices. By checking for potential flaws using a specialised Docker linter, performance and security improvements can be easily identified, saving countless hours of wasted time or security issues in production code.
Otherwise: Mistakenly the Dockerfile creator left Root as the production user, and also used an image from unknown source repository. This could be avoided with with just a simple linter.
Milestones
To maintain this guide and keep it up to date, we are constantly updating and improving the guidelines and best practices with the help of the community. You can follow our milestones and join the working groups if you want to contribute to this project
Translations
All translations are contributed by the community. We will be happy to get any help with either completed, ongoing or new translations!
Completed translations
- Brazilian Portuguese - Courtesy of Marcelo Melo
- Chinese - Courtesy of Matt Jin
- Russian - Courtesy of Alex Ivanov
- Polish - Courtesy of Michal Biesiada
- Japanese - Courtesy of Yuki Ota, Yuta Azumi
- Basque - Courtesy of Ane Diaz de Tuesta & Joxefe Diaz de Tuesta
Translations in progress
- French (Discussion)
- Hebrew (Discussion)
- Korean - Courtesy of Sangbeom Han (Discussion)
- Spanish (Discussion)
- Turkish (Discussion)
Steering Committee
Meet the steering committee members - the people who work together to provide guidance and future direction to the project. In addition, each member of the committee leads a project tracked under our GitHub projects.
Independent Node.js consultant who works with customers in the USA, Europe, and Israel on building large-scale Node.js applications. Many of the best practices above were first published at goldbergyoni.com. Reach Yoni at @goldbergyoni or me@goldbergyoni.com
Full Stack Software Engineer / Developer specializing in Security, DevOps/DevSecOps, and ERP Integrations.
Full Stack Developer who knows how to exit from Vim and loves Architecture, Virtualization and Security.
Contributing
If you've ever wanted to contribute to open source, now is your chance! See the contributing docs for more information.
Contributors β¨
Thanks goes to these wonderful people who have contributed to this repository!
Steering Committee Emeriti
Full Stack Developer & Site Reliability Engineer based in New Zealand, interested in web application security, and architecting and building Node.js applications to perform at global scale.
Independent full-stack developer with a taste for Ops and automation.
Deep specialist in JavaScript and its ecosystem β React, Node.js, TypeScript, GraphQL, MongoDB, pretty much anything that involves JS/JSON in any layer of the system β building products using the web platform for the worldβs most recognized brands. Individual Member of the Node.js Foundation.