fleet/docs/Contributing/getting-started/building-fleet.md

Building Fleet

• Building the code
  • Quickstart
  • Clone and build
  • Details
• Development infrastructure
  • Starting the local development environment
  • Running Fleet using Docker development infrastructure
  • Customizing service ports
  • Running a second Fleet server
• Debugging with Delve debugger

Building the code

Quickstart

Install the dependencies as described in the following sections, then go to Clone and build

macOS

Enable the macOS developer tools:

xcode-select --install

Install Homebrew to manage dependencies, then:

brew install git go node yarn

Ubuntu

Install dependencies:

sudo apt-get install -y git golang make nodejs npm
sudo npm install -g yarn
# Install nvm to manage node versions (apt very out of date) https://github.com/nvm-sh/nvm#install--update-script
curl -o- https://raw.githubusercontent.com/nvm-sh/nvm/v0.40.3/install.sh | bash
# refresh your session before continuing
nvm install v24.10.0

Windows

To install dependencies, we recommend using Chocolatey. Always run Chocolatey in Powershell as an Administrator. Assuming your setup does not include any of our requirements, please run:

choco install nodejs git golang docker make python2 mingw
npm install -g yarn

Note: all packages default to the latest versions. To specify a version, place --version <version-number> after each package. You may also install all packages manually from their websites if you prefer.

After installing the packages, you must use Git Bash to continue with the next section.

If you plan to use WSL on your windows development environment, you can configure Docker to WSL integration by following the steps in Microsoft's WSL Documentation.

Clone and build

The following commands will clone the repository and build the Fleet and fleetctl binaries. When running these commands for the first time, please review the Details section below for additional required dependencies.

git clone https://github.com/fleetdm/fleet.git
cd fleet
make deps
make generate
make

The binaries are now available in ./build/.

Details

To set up a working local development environment, you must have the following toolset:

• Go
• Node.js v24.10.0 and Yarn
  • A specific version of Node.js can be installed using nvm
• GNU Make (probably already installed if you're on macOS/Linux)

Once you have those requirements, check out this Loom video that walks through starting up a local development environment for Fleet.

For a text-based walkthrough, continue through the following steps:

First, you will need to install Fleet's dependencies.

To do this, run the following from the root of the repository:

make deps

Note: If you are using python >= 3.12, you may have to install distutils using pip.

pip install setuptools

or

pip3 install setuptools

When pulling changes, it may be necessary to re-run make deps if a new Go or JavaScript dependency was added.

Generating the packaged JavaScript

To generate all necessary code (bundling JavaScript into Go, etc.), run the following:

make generate

If you are using a Mac computer with Apple Silicon and have not installed Rosetta 2, you will need to do so before running make generate. Otherwise, you may see the following error: Unknown system error -86

/usr/sbin/softwareupdate --install-rosetta --agree-to-license

Automatic rebuilding of the JavaScript bundle

Usually, make generate takes the JavaScript code, bundles it into a single bundle via Webpack, and inlines that bundle into a generated Go source file so that all of the frontend code can be statically compiled into the binary. When you build the code after running make generate, include all of that JavaScript in the binary.

This makes deploying Fleet a dream since you only have to worry about a single static binary. If you are working on frontend code, it is likely that you don't want to have to manually re-run make generate and make build every time you edit JavaScript and CSS in order to see your changes in the browser. Instead of running make generate to solve this problem, before you build the Fleet binary, run the following command:

make generate-dev

Instead of reading the JavaScript from an inline static bundle compiled within the binary, make generate-dev will generate a Go source file which reads the frontend code from disk and run Webpack in "watch mode."

Note that when you run make generate-dev, Webpack will be watching the JavaScript files that were used to generate the bundle so that the process will be long-lived. Depending on your personal workflow, you might want to run this in a background terminal window.

After you run make generate-dev, run make build to build the binary, launch the binary and you'll be able to refresh the browser whenever you edit and save frontend code.

Compiling the Fleet binary

For convenience, Fleet includes a Makefile to build the code:

make

It's unnecessary to use Make to build the code, but using Make allows us to account for cross-platform differences more effectively than the go build tool when writing automated tooling. Use whichever you prefer.

Development infrastructure

The following assumes that you already installed Docker and Docker Compose (installed by default with Docker on macOS and Windows).

Starting the local development environment

To set up a canonical development environment via Docker, run the following from the root of the repository:

docker compose up

Note: you can customize the DB Docker image via the environment variables FLEET_MYSQL_IMAGE and FLEET_MYSQL_PLATFORM. For example:

• To run in macOS M1+, set FLEET_MYSQL_PLATFORM=linux/arm64/v8
• To test with MariaDB, set FLEET_MYSQL_IMAGE to mariadb:10.6 or the like (note MariaDB is not officially supported).

Stopping the local development environment

If you'd like to shut down the virtual infrastructure created by Docker, run the following from the root of the repository:

docker compose down

Setting up the database tables

Once you docker compose up and are running the databases, you can build the code and run the following command to create the database tables:

./build/fleet prepare db --dev

Running Fleet using Docker development infrastructure

To start the Fleet server backed by the Docker development infrastructure, run the Fleet binary as follows:

./build/fleet serve --dev

To develop Fleet's MDM features, you have to pass a --fleet_server_private_key in your fleet serve --dev command. Learn how to generate a private key in Fleet's server configuration documentation.

Customizing service ports

All Docker service ports in docker-compose.yml can be overridden with environment variables. This is useful for avoiding port conflicts or running multiple independent Fleet instances. The defaults match the original hardcoded values, so existing setups are unaffected.

Service	Environment variable(s)	Default
mysql	FLEET_MYSQL_PORT	3306
mysql_test	FLEET_MYSQL_TEST_PORT	3307
mysql_replica_test	FLEET_MYSQL_REPLICA_TEST_PORT	3310
redis	FLEET_REDIS_PORT	6379
s3 (RustFS)	FLEET_S3_PORT, FLEET_S3_CONSOLE_PORT	9000, 9001
mailhog	FLEET_MAILHOG_WEB_PORT, FLEET_MAILHOG_SMTP_PORT	8025, 1025
mailpit	FLEET_MAILPIT_WEB_PORT, FLEET_MAILPIT_SMTP_PORT	8026, 1026
smtp4dev_test	FLEET_SMTP4DEV_WEB_PORT, FLEET_SMTP4DEV_SMTP_PORT	8028, 1027
saml_idp	FLEET_SAML_IDP_HTTP_PORT, FLEET_SAML_IDP_HTTPS_PORT	9080, 9443
localstack	FLEET_LOCALSTACK_PORT, FLEET_LOCALSTACK_LEGACY_PORT	4566, 4571
cadvisor	FLEET_CADVISOR_PORT	5678
prometheus	FLEET_PROMETHEUS_PORT	9090

You can verify resolved ports with:

docker compose config

Or with custom overrides:

FLEET_MYSQL_PORT=3326 docker compose config

Running a second Fleet server

You can run a fully independent second Fleet instance alongside the first by using a separate Docker Compose project with different ports. This is useful for parallelizing work (e.g. one instance for development, another for QA).

# Set ports for the second instance
export FLEET_MYSQL_PORT=3326
export FLEET_REDIS_PORT=6389
export FLEET_S3_PORT=9020
export FLEET_S3_CONSOLE_PORT=9011
export FLEET_SERVER_PORT=8090
export FLEET_S3_CARVES_ENDPOINT_URL=http://localhost:$FLEET_S3_PORT
export FLEET_S3_SOFTWARE_INSTALLERS_ENDPOINT_URL=http://localhost:$FLEET_S3_PORT

# Start containers with a separate compose project
docker compose -p fleet-alt up -d mysql redis s3

# Migrate the database
./build/fleet prepare db \
  --mysql_address=localhost:$FLEET_MYSQL_PORT \
  --mysql_username=root \
  --mysql_password=toor \
  --mysql_database=fleet

# Start the Fleet server
./build/fleet serve \
  --dev --dev_license \
  --server_address=localhost:$FLEET_SERVER_PORT \
  --mysql_address=localhost:$FLEET_MYSQL_PORT \
  --mysql_username=fleet \
  --mysql_password=insecure \
  --mysql_database=fleet \
  --redis_address=localhost:$FLEET_REDIS_PORT \
  --logging_debug

The -p fleet-alt flag gives the second instance separate containers and volumes, so data is fully isolated.

Important: Both instances run on localhost, so browser session cookies are shared between them. Use an incognito window or a different browser profile for the second instance to avoid getting logged out when switching between them.

Note: You would typically run the second instance from a separate clone or git worktree (e.g. ~/fleet2), which also lets you build and run different code versions.

Developing the Fleet UI

When the Fleet server is running, the Fleet UI is accessible by default at https://localhost:8080.

Note that ./build/fleet serve --dev requires the use of make generate-dev because the server will not use bundled assets in this mode. (You may see an error mentioning a template not found when visiting the website otherwise.)

By default, Fleet will try to connect to servers running on default ports on localhost. Depending on your browser's settings, you may have to click through a security warning.

The Fleet UI is developed with Typescript using the React library and SCSS for styling. The source code can be found in the frontend directory.

Debugging with Delve debugger

The Delve Go debugger can be used for debugging the Fleet binary.

Use the following command in place of make and ./build/fleet serve --dev:

dlv debug --build-flags '-tags=full' ./cmd/fleet -- serve --dev

It is important to pass the -tags=full build flag; otherwise, the server will not have access to the asset files.

Attaching a debugger to a running server

You can also run delve in headless mode, which allows you to attach your preferred debugger client and reuse the same session without having to restart the server:

dlv debug --build-flags '-tags=full' --headless \
  --api-version=2 --accept-multiclient --continue \
  --listen=127.0.0.1:61179 ./cmd/fleet -- serve --dev

• If you're using Visual Studio Code, there's a launch configuration in the repo.
• If you're using vim with vimspector, you can use the following config:

{
  "configurations": {
    "Go: Attach to Fleet server": {
      "adapter": "multi-session",
      "variables": {
        "port": 61179,
        "host": "127.0.0.1"
      },
      "configuration": {
        "request": "attach",
        "mode": "remote"
      }
    }
  }
}