Skip to content
Spike Documentation

Install and verify the command-line tools

You need two command-line tools on your own computer. Podman builds and publishes the container images used by your workloads. The Run:ai CLI connects to Spike so you can inspect and control workloads from a terminal.

Before you start

Section titled “Before you start”

The Run:ai CLI can only be installed through the Spike portal after your account has been created and onboarding has started. If you cannot sign in at spike.run.ai, return to Accounts and project access or ask your SCC AI engineer to check the account.

Install eduVPN before testing access from outside the TU/e network. Use your TU/e institutional account and the institute-access profile provided by TU/e.

Install Podman

Section titled “Install Podman”

Follow the official Podman installation instructions for Windows, macOS, or your Linux distribution. Download Podman only from the official project or your operating system’s trusted package repository.

On Windows and macOS, Podman runs Linux containers in a small virtual machine. Create and start it after installing:

Terminal window
podman machine init
podman machine start

Linux normally runs Podman directly and does not need podman machine. Confirm that Podman is ready:

Terminal window
podman version
podman info

If the computer is managed by TU/e and software installation is blocked, ask your local IT support team to install the official package.

Install the Run:ai CLI

Section titled “Install the Run:ai CLI”

The download is provided by your authenticated Run:ai environment, so there is no public Spike-specific installer link to copy ahead of time.

  1. Sign in at spike.run.ai.
  2. Click the Help (?) icon in the top-right corner.
  3. Select Researcher Command Line Interface.
  4. Select the Spike cluster and your computer’s operating system.
  5. Copy the installer command and run it in your terminal. On Windows, download and run the offered installer or binary.
  6. Follow the prompts and keep the recommended defaults unless your SCC AI engineer has given you different settings.

See NVIDIA’s Run:ai CLI installation guide for screenshots, operating-system details, and troubleshooting. Verify the installation in a new terminal:

Terminal window
runai version

Sign in and verify the tools

Section titled “Sign in and verify the tools”

  1. Connect eduVPN when you are outside the TU/e network.

  2. Run runai login. Complete the TU/e sign-in in the browser window that opens, then return to the terminal.

  3. Confirm in the Run:ai portal that the intended project is visible before submitting or changing a workload.

  4. Sign in to Harbor interactively:

    Terminal window
    podman login harbor.spike.tue.nl

  5. Use Podman and Harbor to build and publish a small image before preparing the real workload.