Docker

This guide covers pulling the RLC Pro AI OCI image from the CIQ Depot registry and running it with Docker. On Rocky Linux and RHEL family hosts, Podman is the native runtime; use this guide when your environment standardizes on Docker (CI systems, existing Docker hosts, or developer workstations).

Prerequisites

• A CIQ Portal account with access to the RLC Pro AI product, and your credentials from the Portal access token page.
• Roughly 7 GB of free disk space for the image.

Install Docker

On RLC Pro AI and other Rocky Linux family hosts, install Docker Engine from Docker's upstream repository, following the same procedure as the Rocky Linux Docker gemstone:

sudo dnf config-manager --add-repo https://download.docker.com/linux/rhel/docker-ce.repo
sudo dnf -y install docker-ce docker-ce-cli containerd.io docker-buildx-plugin docker-compose-plugin
sudo systemctl --now enable docker

To let a non-root user run Docker commands, add them to the docker group and log out and back in:

sudo usermod -a -G docker $(whoami)

On non-Rocky platforms (developer laptops, CI images), see the Docker installation documentation instead.

Authenticate to the Depot registry

echo <TOKEN> | docker login depot.ciq.com -u <DEPOT_USERNAME> --password-stdin

Login Succeeded

Passing the token on stdin keeps it out of your shell history and process list.

Pull the image

docker pull depot.ciq.com/rlc-ai-9/rlc-9-ai-oci-images/rlc-pro-ai:9

Confirm the image is present:

docker images depot.ciq.com/rlc-ai-9/rlc-9-ai-oci-images/rlc-pro-ai

Verify the contents

The container carries the same CIQ-built, RPM-managed AI stack as the full operating system. Check the OS identity and the key packages:

docker run --rm depot.ciq.com/rlc-ai-9/rlc-9-ai-oci-images/rlc-pro-ai:9 \
  bash -c "head -4 /etc/os-release; rpm -q python3-torch python3-torchvision"

NAME="Rocky Linux from CIQ"
VARIANT="Pro AI"
VARIANT_ID="pro_ai"
VERSION="9.7 (Blue Onyx)"
python3-torch-2.8.0-12.el9_ciqai.x86_64
python3-torchvision-0.23.0-1.el9_ciqai.x86_64

Run an interactive session

docker run -it --rm depot.ciq.com/rlc-ai-9/rlc-9-ai-oci-images/rlc-pro-ai:9 bash

From there, python3 has the pre-installed PyTorch stack available, exactly as described in the AI Frameworks guide.

Use as a base image

Reference the image in a Dockerfile to build your own application images on top of the validated stack:

FROM depot.ciq.com/rlc-ai-9/rlc-9-ai-oci-images/rlc-pro-ai:9

RUN dnf install -y python3-pip && dnf clean all
COPY requirements.txt /app/
RUN pip3 install -r /app/requirements.txt

GPU access

To give containers access to NVIDIA GPUs, install the NVIDIA Container Toolkit on the Docker host. On an RLC Pro AI host, CIQ delivers it in the rlc-9-supplemental repository:

sudo dnf install nvidia-container-toolkit

Configure the Docker runtime and restart the daemon:

sudo nvidia-ctk runtime configure --runtime=docker
sudo systemctl restart docker

Then attach GPUs with the --gpus flag:

docker run --rm --gpus all \
  depot.ciq.com/rlc-ai-9/rlc-9-ai-oci-images/rlc-pro-ai:9 \
  python3 -c "import torch; print(torch.cuda.is_available())"

See the NVIDIA Container Toolkit documentation for details.

Troubleshooting

authentication required when pulling. The registry rejects unauthenticated pulls. Run docker login depot.ciq.com with your Portal credentials and retry. For other registry issues, see the Depot container guide.

Pull fails partway through. Check free disk space; the image needs roughly 7 GB in Docker's storage location (docker system df shows usage).