Deploy via Web Interface

The Container Manager web interface provides a visual way to deploy and manage services. You can deploy services using a code package with drag-and-drop upload or by providing a Docker image URL.

Access the Container Manager

1. Log in to the Arango Contextual Data Platform web interface.
2. Go to Control Panel in the main navigation sidebar and then click Container Manager.
3. The Container Manager opens with two tabs:
   • Packages: For deploying services from code packages and managing uploaded packages.
   • Containers: For deploying services from Docker images and managing running containers.

Each tab has a Deploy new Service panel on the left and a list of existing services on the right.

Deploy Service from a Code Package

1. Select the Packages tab.
2. In the Deploy new Service panel, drag and drop your .tar.gz file into the upload area, or click to browse and select your file.
3. Enter a File name for your service (e.g., ml-prediction-service).
4. Specify a Version using semantic versioning (e.g., 1.0.0, 2.1.3). This allows you to maintain multiple versions of the same service.
5. Choose the Base Image from the dropdown:
   • py12base: Python 3.12 base runtime
   • py12torch: Python 3.12 with PyTorch
   • py12cugraph: Python 3.12 with cuGraph
   • node22base: Node.js 22 base runtime
6. Define the Service URL Path where your service will be accessible (e.g., my-service).
7. Check Make this a global URL service to make the service accessible globally across all databases. Leave it unchecked for database-specific services.
8. Click Deploy Service. The platform uploads your package, provisions the resources, and starts your service in the Kubernetes cluster.

Update a Code Package Service

To deploy a new version of an existing service:

1. Follow the steps in Deploy Service from a Code Package.
2. Use the same File name as the existing service.
3. Provide a new Version number (e.g., increment from 1.0.0 to 1.1.0).
4. Upload the updated service package.
5. Click Deploy Service.

The new version is deployed alongside the existing version. You can run multiple versions simultaneously. Additionally, you can run multiple instances of the same version, as long as each instance has a different app_instance_name per database. This allows you to deploy the same service version to different databases or create multiple instances within the same database context with unique identifiers.

Deploy Service from a Docker Image

1. Select the Containers tab.
2. In the Deploy new Service panel, enter the Image URL for your Docker image (e.g., docker.io/org/image:tag).
3. Define the Service URL Path where your service will be accessible (e.g., my-service).
4. Check Make this a global URL service to make the service accessible globally across all databases. Leave it unchecked for database-specific services.
5. Click Deploy Container. The platform pulls your Docker image, provisions the resources, and starts your service in the Kubernetes cluster.

Your Docker image must expose an HTTP server (default port: 8000) and handle requests at the root path (/).

Register a Service as an App

If your service serves a user interface (HTML) at the root path (/), you can register it as an App so it becomes available in the platform’s Apps catalog and its UI is rendered embedded in the web interface. This works for both code package and Docker image deployments.

1. Fill in the Deploy new Service panel for a code package (see Deploy Service from a Code Package) or a Docker image (see Deploy Service from a Docker Image).
2. Enable Register as App.
3. Enter a Display name and a Description. These identify your app in the Apps catalog.
4. Click Deploy Service (or Deploy Container).

Once deployed, the service appears in the Packages or Containers list with an Apps badge. You can then open it from the Apps catalog. For details, see Host a UI with Apps.

Stop a Service

To stop a running service:

1. Locate the service in the Running filter of the relevant tab (Packages or Containers).
2. Click the service card to open the detail view.
3. Click Stop Service.
4. Confirm the action.

API Alternative

For programmatic deployment and automation, see Deploy via API.