Worker Installation

Start the AODT Digital Twin server with Docker Compose
View as Markdown

The worker stack starts the AODT Digital Twin server and its supporting services with Docker Compose. Client APIs connect to this stack when talking to a running Digital Twin server.

Deployment types

DeploymentWorker setup
ColocatedIf you completed Client Installation, prerequisites are already installed. Proceed to Start the worker.
Non-colocatedInstall and start the worker on the remote GPU host so the client can connect over the network. Follow the instructions below.

Prerequisites

SSH access to the worker host is required.

Install GPU drivers, Docker, and the NVIDIA Container Toolkit on the worker machine. Follow the steps on the Prerequisites page.

Clone the AODT repository

$git clone https://github.com/NVIDIA/aerial-omniverse-digital-twin.git aodt_1.5.0

This creates an aodt_1.5.0 directory. No additional worker installation steps are required.

Start the worker

From the package root:

$cd aodt_1.5.0
$./worker/up.sh

The script detects the machine IP automatically, then starts the Digital Twin worker together with ClickHouse, MinIO, and Nessie.

Stop the worker

$docker compose -f worker/docker-compose.yml down

Verify a non-colocated deployment

Colocated deployments are verified during Client Installation. To test a non-colocated setup, start the worker on the remote host using the commands above, then follow Verify Installation — non-colocated test from the client machine.

Development mode

To modify worker source code or rebuild from inside the container, see Development Workflow. Development mode is independent of deployment type — both colocated and non-colocated setups support it.

Next steps