Overview

Use Container Management to deploy a specific version of your containerized application to WEDA Nodes, check deployment results, and manage the containers on each device.

Containers provide the application runtime within your Edge AI solution. This chapter covers that part of the lifecycle, alongside the separate management of devices, data and AI model files.

The Full Picture

The Stack defines the application. A Deployment associates a selected revision with its target devices. Registry configurations provide the image registry access settings needed by WEDA Node.

Image and Container

• Image: A packaged application and its runtime dependencies, stored in an image registry.
• Container: An instance of an image on a device. It can be running, paused or stopped.

Choose an image that supports your device's CPU architecture, hardware and runtime requirements. The image may run your own application, an AI inference service, or an integration component such as a WEDA SubNode SDK application.

Note: Plan device storage

Allow disk space for image layers, each container's writable data, persistent volumes and logs. Reserve headroom for pulling and updating images while versions you still need remain on the device. Check the filesystems used by Docker and any host data mounts, and review the selected application's ACC listing or version-specific README for its requirements.

Note: AI model files and container images

A model file is separate from the container image that provides its runtime. Use AI Model Management to manage model files and their deployment. The application must be configured to read and use the deployed model.

Registry

An image registry stores container images. WEDA Node pulls those images when deploying the application.

A registry configuration is the record you manage in WEDA for a registry's address, type and access credentials. It belongs to an Org and supplies the settings required to access the registry during deployment.

Keep these responsibilities separate:

• The registry stores the images. Examples include Docker Hub, AWS ECR, Azure ACR and a self-hosted registry such as Harbor.
• WEDA stores the access configuration. Provide the credentials required by the selected registry type.
• WEDA Node pulls the images. The target device needs network access to the registry.

Configure these settings in Registry Management.

Stack and Revision

A Container Stack is an application definition based on Docker Compose. It describes one or more services, including the images, ports, volumes and environment settings they need.

Resource	What it represents
Stack	The application definition you manage in WEDA, identified by its Stack name and ID.
Revision	A saved version of the Stack's Compose content, environment values and description. Select a revision when deploying.

The Compose file references images by their image names. Registry access settings are managed separately from the Stack definition.

Updating a Stack creates a new revision. You can inspect its revisions, compare configurations and select the revision to deploy.

Note: Updating a Stack does not update its deployments

Creating a new revision leaves existing deployments unchanged. Deploy the selected revision to the devices you want to update.

Running a single container

A Stack can contain just one service. Use the same Stack and Revision model for a single-container application.

Create and manage these definitions in Container Stack Management.

Deployment

A Deployment applies a selected Stack revision to one or more WEDA Nodes. WEDA Node receives the configuration, pulls the referenced images from their registries and starts the containers.

When managing a deployment, distinguish:

• The requested revision: Which application configuration you selected for the device.
• The deployment result: Whether the device completed the deployment operation.
• The runtime state: Whether the deployed containers are currently running, paused or stopped.

Check the device's reported result and runtime state after deployment. Submitting the request alone does not confirm that the application is running. Different target devices can run different revisions of the same Stack.

Commands

Container commands control an existing deployment's runtime state without creating a new Stack revision.

Command	Purpose
start	Start stopped containers.
stop	Stop running containers.
pause	Pause running containers.
resume	Resume paused containers.
restart	Restart containers.

For example, you can stop a deployed workload during maintenance and start it again afterward. Check the command's execution result and the device's reported runtime state, rather than relying on an earlier successful deployment result.

Next Steps

The task guides combine WEDA Console operations with API details where applicable:

Guide	Use it to
Registry Management	Configure image registry access for an Org.
Container Stack Management	Create application definitions and manage their revisions.
Container Deployment & Commands	Deploy a revision, check results and control deployed containers.

For your first deployment through WEDA Console, follow Deploy Your First Application.