Bluefin Developer Experience (
bluefin-dx) is a dedicated developer image with bundled tools. Unlike traditional Linux systems, the operating system and developer environment are explicitly and purposely decoupled.
This means that tooling is not installed on the host, and is instead containerized, in a virtual machine, or scoped to the user’s home directory.
It is designed to meet the following use cases:
- Endeavors to be the world’s most powerful cloud native developer environment
- Full virtualization support
- Provide fleet management of Bluefin and other Linux systems via Cockpit (Incomplete)
You can rebase to
bluefin-dx by using the following command:
just devmodeto enable or disable the dx mode:
Like all Universal Blue images, switching is atomic, allowing for clean switching between modes depending on the use case.
Bluefin offers four “default” developer workflows. Since development is not dependant on the operating system image, you can use whatever you want.
- Dev Containers Documentation - you can skip most of the installation instructions and go directly to the tutorial
- Dev Containers Specification
- Beginner’s Series to: Dev Containers - great introductory tutorial from the VS Code YouTube channel
just distrobox-brew will install the upstream container as a distrobox on your system.
- Follow Quickstart VS Code to set up your envitronment.
just nix-devboxto get started
just nix-devbox-globalto install a global profile
- Built-in Ubuntu user space
- Ctrl-Alt-u - will launch an Ubuntu image inside a terminal via Distrobox and your home directory will be transparently mounted for the Ubuntu image to access
- A BlackBox terminal is used just for this configuration
- Use this container for your typical CLI needs or to install software that is not available via Flatpak or Fedora
- Optional ubuntu-toolbox image with Python, and other convenience development tools.
just distrobox-bluefinto get started. To configure
justfollow the guide.
just distrobox-universalto get started
just assembleshortcut to declaratively build distroboxes defined in
- Refer to the Distrobox documentation for more information on using and configuring custom images
- GNOME Terminal - Ctrl-Alt-t - will launch a host-level GNOME Terminal if you need to do host-level things in Fedora (you shouldn’t need to do much).
just jetbrains-toolbox will fetch and install the JetBrains Toolbox application, which will manage the installation of the JetBrains set of tools. This application will handle installation, removal, and upgrade of the JetBrains products, and is handled completely in your home directory, independent of the operating system image.
- kind - Run a Kubernetes cluster on your machine. Run
kind create clusteron the host to get started!
- kubectl - Administer Kubernetes Clusters
- helm, ko, flux, minio-client – if it’s an incubated project we intend to add it where appropriate
The most current Docker Engine is included by default. Use
just docker to turn it on, this will turn on the service and add your users to the
docker group. You will then need to log out and back into your desktop.
To configure Visual Studio Code to use Docker go to the extensions section, and click the gear on the bottom right of the remote containers extension to configure it:
Then change the following defaults to point to
- virt-manager and associated tooling (KVM, qemu)
- Incus provides system containers
Bluefin includes a NGC container that includes the latest stable PyTorch from Nvidia on top of Ubuntu LTS. It includes
pip packages are commonly used but not comprehensive. Only
pip is used and
conda are not currently tested.
You must already be using
bluefin-dx-nvidia as it’s meant for those GPUs and has nvidia container toolkit support. If you want to test before downloading the larger nvidia container run
inside a regular Ubuntu box (
distrobox create --nvidia -i ubuntu) to check that everything works correctly.
To create the box use the following command. Note that this container is large (20+ GB!):
To enter the working environment:
distrobox enter mlbox
Then the init-hooks will run once. After which, you should be able to run:
To check if GPUs are seen and enter the python repo run:
Various tests can be run to test a transformers inference or training job, or to git clone a pytorch benchmarks repo and run single or multi gpu commands: E.g. to test multi-gpu setup on two 3090s:
git clone https://github.com/aime-team/pytorch-benchmarks.git cd pytorch-benchmarks python3 main.py --num_gpus 2 --compile --model bert-large-uncased --data_name squad --global_batch_size 24
On other operating systems, use this .ini file and run:
distrobox assemble create --file /path/to/your/mlbox.ini
- Cockpit for local and remote management
- A collection of well curated monospace fonts
- systemd shutdown timers adjusted to 15 seconds
- Tailscale for VPN
- Just task runner for post-install automation tasks. Check out our documentation for more information on using and customizing just.
zshavailable as optional shells, use
just zshand follow the prompts to configure them