Skip to main content

Run the manager and JupyterLab using Docker

The AI Unlimited manager orchestrates the engine's deployment and includes a web-based user interface for monitoring projects. And the manager is where you'll set up AI Unlimited.

You'll use JupyterLab to explore and analyze data.

You'll use Docker Compose to run the AI Unlimited manager and JupyterLab, with the AI Unlimited Jupyter Kernel, locally in containers.

Tip

For installation support, email the support team or ask the community.

Prerequisites

  • A pay-as-you-go AWS or Azure account on which to deploy the engine from a Jupyter notebook
  • A GitHub or GitLab account to host each project repository for authenticating users and storing project information
  • Your object storage, where your Amazon or ADLS Gen2 data lake resides
  • Docker installed on your computer

Set configuration file locations

  1. Optionally, set the AI_UNLIMITED_HOME environment variable to the directory in which to store the manager's configuration and data files. Make sure the directory exists, and that appropriate permission is granted. The default location is ./volumes/ai-unlimited.

    Local locationContainer locationUsage
    $AI_UNLIMITED_HOME/etc/tdStores data and configuration
    Tip

    Learn about AWS or Azure environment variables.

  2. Optionally, set the JUPYTER_HOME environment variable to the directory in which to store the JupyterLab configuration files. The default location is ~/.jupyter.

Clone the repository

The deployments/docker folder in the AI Unlimited GitHub repository provided by Teradata includes these files that you'll need to run the manager and JupyterLab:

  • [AWS or Azure]-credentials-env-vars.yaml
  • ai-unlimited.yaml
  • jupyter.yaml

Clone the repository.

Pass your cloud service provider credentials to Docker

Note

You can pass the credentials two ways:

  • Use [AWS or Azure]-credentials-env-vars.yaml, which contains environment variables for storing your credentials.
  • Use a local volume containing your credentials.

See both methods in the Jupyter and AI Unlimited section of Deploy with Docker Compose in the Teradata AI Unlimited GitHub repository.

This QuickStart uses the first method.

  1. Copy these environment variables from your cloud service provider's console.

    AWS_ACCESS_KEY_ID, AWS_SECRET_ACCESS_KEY, and AWS_SESSION_TOKEN

  2. Go to the directory where [AWS or Azure]-credentials-env-vars.yaml is located and update the file's environment variable values.

Start the manager and JupyterLab

  1. From the directory where [AWS or Azure]-credentials-env-vars.yaml, ai-unlimited.yaml, and jupyter.yaml are located, start the manager and JupyterLab.

    Note

    The -d flag in the command is optional.

    The command downloads and starts the manager and JupyterLab containers.

  2. To retrieve the Jupyter token, list the currently running containers.

    And identify the name of the JupyterLab container.

    Then search for occurrences of the string 'Token' in the container's logs.

Verify access

When the manager is ready, you can access it at http://localhost:3000.

When JupyterLab is ready, you can access it at http://localhost:8888, and enter the token.

What's next

Create an OAuth app to allow authentication between AI Unlimited and your Git provider account.

Also of interest