Installation¶

Install the latest l8k binary, then select the target Network Operator release with --network-operator-release. A single current l8k binary carries the release catalog for older supported Network Operator lines.

Prerequisites¶

Linux or macOS on amd64 or arm64.
Kubernetes credentials for discovery, deployment, validation, and sosreport collection.
kubectl is recommended for inspection and troubleshooting.
curl for the install script, Homebrew for the formula, or Docker/Podman for the container method.

A pre-installed Network Operator and NFD are not required for discovery. l8k deploy can install the selected Network Operator Helm chart from the generated values.yaml.

Install Script¶

curl -fsSL https://raw.githubusercontent.com/NVIDIA/k8s-launch-kit/main/scripts/install.sh | sh

Pin a release:

L8K_VERSION=v26.7.0-beta.4 \
  sh -c "$(curl -fsSL https://raw.githubusercontent.com/NVIDIA/k8s-launch-kit/main/scripts/install.sh)"

Use a custom destination:

curl -fsSL https://raw.githubusercontent.com/NVIDIA/k8s-launch-kit/main/scripts/install.sh | \
  sh -s -- -d "$HOME/.local"

Homebrew¶

brew tap nvidia/l8k https://github.com/NVIDIA/k8s-launch-kit
brew install l8k

Container¶

Set L8K_VERSION to a published release tag and pull the image:

export L8K_VERSION=vX.Y.Z
docker pull nvcr.io/nvidia/cloud-native/k8s-launch-kit:${L8K_VERSION}

Define a shell function that mounts the kubeconfig and current working directory:

l8k() {
  docker run --rm --network host \
    -v "$HOME/.kube:/root/.kube:ro" \
    -v "$PWD:/work" \
    -w /work \
    "nvcr.io/nvidia/cloud-native/k8s-launch-kit:${L8K_VERSION}" \
    "$@"
}

The generated configuration, manifests, and reports remain in the current directory. Ensure the Kubernetes API server address in the mounted kubeconfig is reachable from the container runtime.

Build From Source¶

git clone https://github.com/NVIDIA/k8s-launch-kit.git
cd k8s-launch-kit
make build

The binary is written to build/l8k.

For a local install:

make install

For development, symlink the source-tree assets instead of copying them:

make dev-install

Installed Assets¶

The install places the binary on PATH. The default configuration and bundled topology presets are embedded in the binary. Use --config-dir when you need a filesystem override:

/etc/l8k/
|-- l8k-config.yaml
`-- presets/
    `-- <preset-name>/topology.yaml

l8k discover --config-dir /etc/l8k --kubeconfig ~/.kube/config
l8k preset list --config-dir /etc/l8k

--user-config <file> has higher precedence than --config-dir/l8k-config.yaml. A presets/ directory in --config-dir replaces the embedded preset catalog instead of merging with it.

Script, Homebrew, and source installs also place the profile templates under <prefix>/share/l8k/profiles/. Existing l8k-config.yaml and presets/ overrides under the share directory are preserved during upgrades and are selected only when passed through --config-dir.

Verify¶

l8k version
l8k schema | jq '.supportedNetworkOperatorReleases'

Uninstall¶

Install script:

curl -fsSL https://raw.githubusercontent.com/NVIDIA/k8s-launch-kit/main/scripts/install.sh | \
  sh -s -- --uninstall

Use -d <prefix> with --uninstall if a custom install destination was used.

Homebrew:

brew uninstall l8k
brew untap nvidia/l8k

Container:

docker image rm "nvcr.io/nvidia/cloud-native/k8s-launch-kit:${L8K_VERSION}"