On This Page

在 GitHub 上编辑本页
  • Repo
  • 文档
  • Getting Started
  • Add to Existing Monorepo

Add Turborepo to your existing monorepo

turbo works with Yarn, npm, and pnpm on the following operating systems:

  • macOS darwin 64-bit (Intel), ARM 64-bit (Apple Silicon)
  • Linux 32-bit, 64-bit, ARM, ARM 64-bit, MIPS 64-bit Little Endian, PowerPC 64-bit Little Endian, IBM Z 64-bit Big Endian
  • Windows 32-bit, 64-bit, ARM 64-bit
  • FreeBSD 64-bit, ARM 64-bit
  • NetBSD AMD64
  • Android ARM 64-bit

Configure workspaces

turbo is built on top of Workspaces, a way of managing multiple packages from within a single monorepo package. Turborepo is compatible with the workspace implementations from all package managers. For more information on managing your Turborepo workspaces, see the Workspaces documentation.

You can configure workspaces any way you want, but a common folder structure example is keeping applications in the /apps folder and packages in the /packages folder. The configuration of these folders is different for each package manager.

Specify your workspaces in your monorepo's root package.json file:

package.json
{
  "workspaces": ["packages/*", "apps/*"]
}

After configuring your workspaces, re-run your package manager's install command.

Note: Nested workspaces are not supported. As package names are required to be unique, moving each package to be a child of the monorepo's root package should meet your needs.

Install turbo

Add turbo as a development dependency at the root of your monorepo.

npm install turbo -D

The turbo package is a shell script that will install the proper turbo-<os>-<arch> package for your operating system and architecture.

Note: Linux builds of turbo link against glibc. For Alpine Docker environments, you will need to ensure libc6-compat is installed as well, via RUN apk add --no-cache libc6-compat

Create turbo.json

In the root of your monorepo, create an empty file named turbo.json. This will hold the configuration for Turborepo.

./turbo.json
{
  "$schema": "https://turbo.build/schema.json"
}

Create a pipeline

To define your monorepo's task dependency graph, use the pipeline key in the turbo.json configuration file at the root of monorepo. turbo interprets this configuration to optimally schedule, execute, and cache the outputs of each of the package.json scripts defined in your workspaces.

Each key in the pipeline object is the name of a package.json script that can be executed by turbo run. You can specify its dependencies with the dependsOn key inside it as well as some other options related to caching. For more information on configuring your pipeline, see the Pipelines documentation.

Workspaces that do not have the specified script defined in their package.json's list of scripts will be ignored by turbo.

./turbo.json
{
  "$schema": "https://turbo.build/schema.json",
  "pipeline": {
    "build": {
      // A package's `build` script depends on that package's
      // dependencies and devDependencies
      // `build` tasks  being completed first
      // (the `^` symbol signifies `upstream`).
      "dependsOn": ["^build"],
      // note: output globs are relative to each package's `package.json`
      // (and not the monorepo root)
      "outputs": [".next/**"]
    },
    "test": {
      // A package's `test` script depends on that package's
      // own `build` script being completed first.
      "dependsOn": ["build"],
      "outputs": [],
      // A package's `test` script should only be rerun when
      // either a `.tsx` or `.ts` file has changed in `src` or `test` folders.
      "inputs": ["src/**/*.tsx", "src/**/*.ts", "test/**/*.ts", "test/**/*.tsx"]
    },
    "lint": {
      // A package's `lint` script has no dependencies and
      // can be run whenever. It also has no filesystem outputs.
      "outputs": []
    },
    "deploy": {
      // A package's `deploy` script depends on the `build`,
      // `test`, and `lint` scripts of the same package
      // being completed. It also has no filesystem outputs.
      "dependsOn": ["build", "test", "lint"],
      "outputs": []
    }
  }
}

The rough execution order for a given package is based on the dependsOn keys:

  1. build once its upstream dependencies have run their build commands
  2. test once its own build command is finished and has no filesystem outputs (just logs) within a package
  3. lint runs in an arbitrary order as it has no upstream dependencies
  4. deploy once its own build, test, and lint commands have finished.

After execution, the full pipline can run:

npx turbo run build test lint deploy

turbo will then schedule the execution of each task(s) to optimize usage of the machine's resources.

Edit .gitignore

Add .turbo to your .gitignore file. The CLI uses these folders for logs and certain task outputs.

+ .turbo

Make sure that your task artifacts, the files and folders you want cached, are also included in your .gitignore.

+ build/**
+ dist/**
+ .next/**

Re-run your npm client's install command to check your configuration.

Create package.json scripts

Add or update scripts in your monorepo's root package.json file and have them delegate to turbo.

./package.json
{
  "scripts": {
    "build": "turbo run build",
    "test": "turbo run test",
    "lint": "turbo run lint",
    "dev": "turbo run dev"
  }
}

Build your monorepo

npm run build

Depending on your monorepo setup, some artifacts might already be caching properly. In the next sections, we'll show how turbo works, how scope works, and then how to get caching working after that.

Configure Remote Caching

A major key 🔑 to Turborepo's speed is that it is both lazy and efficient—it does the least amount of work possible and it tries to never redo work that's already been done before.

At the moment, Turborepo caches your tasks on your local filesystem (i.e. "single-player mode," if you will). However, what if there was a way to take advantage of the computational work done by your teammates or your CI (i.e. "co-op multiplayer mode")? What if there was a way to teleport and share a single cache across machines? Almost like a "Dropbox" for your Turborepo cache.

Remote Caching has entered the chat.

Turborepo can use a technique known as Remote Caching to share cache artifacts across machines for an additional speed boost.

Remote Caching is a powerful feature of Turborepo, but with great power comes great responsibility. Make sure you are caching correctly first and double check handling of environment variables. Please also remember Turborepo treats logs as artifacts, so be aware of what you are printing to the console.

Using Remote Caching for Local development

Turborepo uses Vercel as its default remote caching provider. If you want to link your local turborepo to your Remote Cache you can authenticate the Turborepo CLI with your Vercel account:

npx turbo login

Then, link your turborepo to your remote cache:

npx turbo link

Once enabled, make some changes to a package or application you are currently caching and run tasks against it with turbo run. Your cache artifacts will now be stored locally and in your Remote Cache. To verify that this worked, delete your local Turborepo cache:

rm -rf ./node_modules/.cache/turbo

Run the same build again. If things are working properly, turbo should not execute tasks locally, but rather download both the logs and artifacts from your Remote Cache and replay them back to you.

Note: When connecting to an sso-enabled Vercel team, you must provide your Team slug as an argument to npx turbo login.

npx turbo login --sso-team=<team-slug>

Next Steps

You're now up and running with Turborepo, but there are still a few things to do:

Create a New MonorepoCaching Tasks