fleet/tools/tuf/test/README.md
Lucas Manuel Rodriguez fa71b582d1
Add -s -w when building fleetd components to remove debugging information and reduce binary sizes (#43260)
Resolves #43259.

- [x] Changes file added for user-visible changes in `changes/`,
`orbit/changes/` or `ee/fleetd-chrome/changes`.
See [Changes
files](https://github.com/fleetdm/fleet/blob/main/docs/Contributing/guides/committing-changes.md#changes-files)
for more information.

## Testing

- [x] QA'd all new/changed functionality manually

## fleetd/orbit/Fleet Desktop

- [x] Verified compatibility with the latest released version of Fleet
(see [Must
rule](https://github.com/fleetdm/fleet/blob/main/docs/Contributing/workflows/fleetd-development-and-release-strategy.md))
- [x] Verified that fleetd runs on macOS, Linux and Windows
- [x] Verified auto-update works from the released version of component
to the new version (see [tools/tuf/test](../tools/tuf/test/README.md))
2026-04-08 18:30:05 -03:00

8 KiB

Testing TUF

Scripts in this directory aim to ease the testing of Orbit and the TUF system.

WARNING: All of these scripts are for testing only, they are not safe for production use.

Setup

  1. The script is executed on a macOS host.
  2. Fleet server also running on the same macOS host.
  3. All VMs (and the macOS host itself) are configured to resolve host.docker.internal to the macOS host IP (by modifying their hosts file).
  4. The hosts are running on the same GOARCH as the macOS host. If not, you can set the GOARCH environment variable to compile for the desired architecture. For example: GOARCH=amd64

PS: We use host.docker.internal because the testing certificate ./tools/osquery/fleet.crt has such hostname (and localhost) defined as SANs.

PPS: Make sure you set the macOSX deployment target to the lowest macOS version you intend to support. See Troubleshooting for more details.

Run

The main.sh creates and runs the TUF repository and optionally generate the installers (GENERATE_PKGS):

SYSTEMS="macos windows linux linux-arm64 windows-arm64" \
PKG_FLEET_URL=https://localhost:8080 \
PKG_TUF_URL=http://localhost:8081 \
DEB_FLEET_URL=https://host.docker.internal:8080 \
DEB_TUF_URL=http://host.docker.internal:8081 \
RPM_FLEET_URL=https://host.docker.internal:8080 \
RPM_TUF_URL=http://host.docker.internal:8081 \
MSI_FLEET_URL=https://host.docker.internal:8080 \
MSI_TUF_URL=http://host.docker.internal:8081 \
PKG_TAR_ZST_FLEET_URL=https://host.docker.internal:8080 \
PKG_TAR_ZST_TUF_URL=http://host.docker.internal:8081 \
GENERATE_PKG=1 \
GENERATE_DEB=1 \
GENERATE_DEB_ARM64=1 \
GENERATE_RPM=1 \
GENERATE_RPM_ARM64=1 \
GENERATE_MSI=1 \
GENERATE_MSI_ARM64=1 \
GENERATE_PKG_TAR_ZST=1 \
GENERATE_PKG_TAR_ZST_ARM64=1 \
ENROLL_SECRET=6/EzU/+jPkxfTamWnRv1+IJsO4T9Etju \
FLEET_DESKTOP=1 \
DEBUG=1 \
./tools/tuf/test/main.sh

Separate *_FLEET_URL and *_TUF_URL variables are defined for each package type to support different setups.

To publish test extensions you can set comma-separated executable paths in the {MACOS|WINDOWS|LINUX}_TEST_EXTENSIONS environment variables: Here's a sample to use the hello_world and hello_mars test extensions:

# Build `hello_word` and `hello_mars` test extensions.
./tools/test_extensions/hello_world/build.sh

[...]
MACOS_TEST_EXTENSIONS="./tools/test_extensions/hello_world/macos/hello_world_macos.ext,./tools/test_extensions/hello_world/macos/hello_mars_macos.ext" \
WINDOWS_TEST_EXTENSIONS="./tools/test_extensions/hello_world/windows/hello_world_windows.ext.exe,./tools/test_extensions/hello_world/windows/hello_mars_windows.ext.exe" \
WINDOWS_ARM64_TEST_EXTENSIONS="./tools/test_extensions/hello_world/windows-arm64/hello_world_windows_arm64.ext.exe,./tools/test_extensions/hello_world/windows-arm64/hello_mars_windows_arm64.ext.exe" \
LINUX_TEST_EXTENSIONS="./tools/test_extensions/hello_world/linux/hello_world_linux.ext,./tools/test_extensions/hello_world/linux/hello_mars_linux.ext" \
LINUX_ARM64_TEST_EXTENSIONS="./tools/test_extensions/hello_world/linux-arm64/hello_world_linux_arm64.ext,./tools/test_extensions/hello_world/linux-arm64/hello_mars_linux_arm64.ext" \
[...]
./tools/tuf/test/main.sh

To build for a specific architecture, you can pass the GOARCH environment variable:

[...]
# defaults to amd64
GOARCH=arm64 \
[...]
./tools/tuf/test/main.sh

To include Escrow Buddy, Nudge, or Swift Dialog on the TUF repository you can use the following variables:

[...]
ESCROW_BUDDY=1 \
NUDGE=1 \
SWIFT_DIALOG=1 \
[...]
./tools/tuf/test/main.sh

Test fleetd with expired signatures on a TUF repository

To generate a TUF repository with shorter expiration time for roles you can set the following environment variables:

[...]
KEY_EXPIRATION_DURATION=5m \
TARGETS_EXPIRATION_DURATION=5m \
SNAPSHOT_EXPIRATION_DURATION=5m \
TIMESTAMP_EXPIRATION_DURATION=5m \
[...]
./tools/tuf/test/main.sh

NOTE: The duration has to be enough time to generate the packages (otherwise the fleetctl package command will fail).

KEY_EXPIRATION_DURATION is used to set the expiration of the root.json signature.

Add new updates

To add new updates (osqueryd or orbit), use push_target.sh.

E.g. to add a new version of orbit for Windows:

source ./tools/tuf/test/load_orbit_version_vars.sh

# Compile a new version of Orbit:
GOOS=windows GOARCH=amd64 go build \
    -o orbit-windows.exe \
    -ldflags="-s -w -X github.com/fleetdm/fleet/v4/orbit/pkg/build.Version=$ORBIT_VERSION \
    -X github.com/fleetdm/fleet/v4/orbit/pkg/build.Commit=$ORBIT_COMMIT" \
    ./orbit/cmd/orbit

# Push the compiled Orbit as a new version
./tools/tuf/test/push_target.sh windows orbit orbit-windows.exe $ORBIT_VERSION

If the script was executed on a macOS host, the Orbit binary will be a universal binary. To push updates you can do:

source ./tools/tuf/test/load_orbit_version_vars.sh

# Compile a universal binary of Orbit:
CGO_ENABLED=1 \
ORBIT_VERSION=$ORBIT_VERSION \
ORBIT_COMMIT=$ORBIT_COMMIT \
ORBIT_BINARY_PATH="orbit-macos" \
go run ./orbit/tools/build/build.go

# Push the compiled Orbit as a new version
./tools/tuf/test/push_target.sh macos orbit orbit-macos $ORBIT_VERSION

E.g. to add a new version of osqueryd for macOS:

# Generate osqueryd app bundle.
make osqueryd-app-tar-gz version=5.5.1 out-path=.

# Push the osqueryd target as a new version
./tools/tuf/test/push_target.sh macos-app osqueryd osqueryd.app.tar.gz 5.5.1

NOTE: Contributors on macOS with Apple silicon ran into issues running osqueryd downloaded from GitHub. Until this issue is root caused, the workaround is to download osqueryd from Fleet's TUF.

E.g. to add a new version of desktop for macOS:

source ./tools/tuf/test/load_orbit_version_vars.sh

# Compile a new version of fleet-desktop
FLEET_DESKTOP_VERSION=$ORBIT_VERSION make desktop-app-tar-gz

# Push the desktop target as a new version
./tools/tuf/test/push_target.sh macos desktop desktop.app.tar.gz $ORBIT_VERSION

Troubleshooting

Fleet Desktop Startup Issue on macOS

When running Fleet Desktop on an older macOS version than it was compiled on, Orbit may not launch it due to an error:

_LSOpenURLsWithCompletionHandler() failed with error -10825

Solution: Set the MACOSX_DEPLOYMENT_TARGET environment variable to the lowest macOS version you intend to support:

export MACOSX_DEPLOYMENT_TARGET=13 # replace '13' with your target macOS version

Issue generating linux-arm64 packages when running Docker Desktop on macOS using Apple Silicon

When running Docker Desktop on macOS using Apple Silicon, enrollment packages for ARM Linux may fail to generate and you may see a warning similar to:

WARNING: The requested image's platform (linux/amd64) does not match the detected host platform (linux/arm64/v8) and no specific platform was requested
[...]
/usr/local/go/pkg/tool/linux_amd64/compile: signal: illegal instruction
make: *** [desktop-linux] Error 1

Solution: In Docker Desktop go to Settings >> General >> Virtual Machine Options and choose the "Docker VMM (BETA)" option. Restart Docker Desktop.

Running without ssl

If you decide that you want to run your local fleet server with the --server_tls=false flag you will need to modify a few ENV variables when running the ./tools/tuf/test/main.sh file.

+ INSECURE=1 \

+ PKG_FLEET_URL=http://localhost:8080 \
- PKG_FLEET_URL=https://localhost:8080 \

+ DEB_FLEET_URL=http://host.docker.internal:8080 \
- DEB_FLEET_URL=https://host.docker.internal:8080 \

+ RPM_FLEET_URL=http://host.docker.internal:8080 \
- RPM_FLEET_URL=https://host.docker.internal:8080 \

+ MSI_FLEET_URL=http://host.docker.internal:8080 \
- MSI_FLEET_URL=https://host.docker.internal:8080 \

+ PKG_TAR_ZST_FLEET_URL=http://host.docker.internal:8080 \
- PKG_TAR_ZST_FLEET_URL=https://host.docker.internal:8080 \

These flags change the way tools/tuf/test/gen_pkgs.sh builds the binaries to properly support a local server not running ssl.