Skip to main content

Troubleshooting GPU Visibility in Docker Containers

If your GPU is visible on the host using nvidia-smi but not inside a Docker container, follow these steps to resolve the issue:

  1. Check Docker Compose Configuration: Ensure your docker-compose.yml file includes the count: all line under the deploy section. This allows Docker to access all available GPUs.

    version: '3.9'
    services:
      your_service_name:
        image: your_image_name
        deploy:
          resources:
            reservations:
              devices:
                - driver: nvidia
                  count: all
                  capabilities: [gpu]

  2. Obtain New Docker Compose File: If your docker-compose.yml file does not include the count: all line, obtain a new docker-compose.yml file from the Comprimato Customer Dashboard. Follow the instructions provided to download the updated file from the Run Docker Container guide.

  3. Verify NVIDIA Container Toolkit Installation: Make sure the NVIDIA Container Toolkit is installed on your host machine. You can follow the installation guide in your platform's preparation guide.

  4. Check NVIDIA-SMI Inside Container: After starting the container, run nvidia-smi inside the container to verify GPU visibility.

docker exec -it transcoder0 nvidia-smi

By following these steps, you should be able to resolve the issue of GPU visibility inside your Docker container.