如何在docker-compose.yml中让容器内OpenGL调用NVIDIA GPU

AI 小北

2026-6-25

Docker Compose配置无法让ROS Noetic容器调用NVIDIA GPU的问题与解决

问题背景

使用Dockerfile和docker-compose构建ROS Noetic容器后，容器内OpenGL始终使用llvmpipe软件渲染，无法调用NVIDIA GPU；但直接通过docker run命令启动的容器可以正常调用GPU，需要修正docker-compose配置。

环境信息

文件夹结构

noetic
├── .devcontainer
│   ├── devcontainer.json
│   ├── docker-compose.yml
│   ├── Dockerfile
│   ├── .p10k.zsh
│   ├── powerlevel10k
│   ├── sources.list
│   └── .zshrc
├── .dockerignore
└── workspace

Dockerfile内容

# base image from noetic
FROM osrf/ros:noetic-desktop-full

RUN mv /etc/apt/sources.list /etc/apt/sources.list.bak
COPY .devcontainer/sources.list /etc/apt/

RUN apt update -yq && apt upgrade -yq && \
apt install -y curl sudo zsh zsh-autosuggestions zsh-syntax-highlighting

RUN chsh -s /bin/zsh \
    && echo "source /opt/ros/noetic/setup.bash" >> /root/.bashrc \
    && echo "source /opt/ros/noetic/setup.zsh" >> /root/.zshrc \
    && echo "export net.ipv4.ip_forward=1" >> /etc/sysctl.conf \
    && /bin/bash -c "source /root/.bashrc" \
    && /bin/zsh -c "source /root/.zshrc" \
    && rm -rf /var/lib/apt/lists/*

ARG user=noetic

RUN useradd --create-home --no-log-init --shell /bin/zsh ${user} \ 
    && adduser ${user} sudo \ 
    && echo "${user}:1" | chpasswd \ 
    && usermod -u 1000 ${user} && usermod -G 1000 ${user} \ 
    && echo "%${user} ALL=(ALL:ALL) ALL" >> /etc/sudoers \ 
    && echo "%${user} ALL=(ALL) NOPASSWD:ALL" >> /etc/sudoers \
    && mkdir -p /home/${user}/software/powerlevel10k

COPY .devcontainer/powerlevel10k /home/${user}/software/powerlevel10k

COPY .devcontainer/.zshrc /home/${user}
COPY .devcontainer/.p10k.zsh /home/${user}
RUN echo "source /home/${user}/software/powerlevel10k/powerlevel10k.zsh-theme" >> /home/${user}/.zshrc \
    && echo "source /home/${user}/.zshrc" >> /root/.zshrc

ENV LANG=en_US.UTF-8 LANGUAGE=en_US:en LC_ALL=en_US.UTF-8

HEALTHCHECK --interval=600s --timeout=20s \
  CMD curl -fs http://localhost/ || exit 1

USER ${user}

Docker Compose版本

docker-compose版本为1.28.0，build d02a7b1a

当前docker-compose.yml内容

version: '3.7'

services:
  noetic:
    container_name: noetic
    build:
      context: ../
      dockerfile: .devcontainer/Dockerfile
    # image: osrf/ros:noetic-desktop
    # hostname: noetic_host
    privileged: true
    network_mode: "host"
    command: /bin/bash
    volumes:
      - /dev:/dev
      # Add more volumes here if needed
      - ../workspace:/workspace
      - /tmp/.X11-unix:/tmp/.X11-unix
    environment:
      - NVIDIA_VISIBLE_DEVICES=all
      - TZ=Asia/Shanghai
      - xpack.monitoring.enabled=false
      - xpack.watcher.enabled=false
      - DISPLAY=$DISPLAY
      - GDK_SCALE
      - GDK_DPI_SCALE
    deploy:
      resources:
        limits:
          cpus: '0.70'
          memory: 8G
        reservations:
          devices:
            - driver: nvidia
              count: all
              capabilities: [gpu]
    healthcheck:
      test: ["CMD", "curl", "-f", "http://localhost"]
      interval: 10m
      timeout: 20s
      retries: 3
    logging:
      driver: "json-file"
      options:
        max-size: "50M"
        max-file: "10"
    tty: true

问题现象

本地主机nvidia-smi输出：

+---------------------------------------------------------------------------------------+
| NVIDIA-SMI 535.161.07             Driver Version: 535.161.07   CUDA Version: 12.2     |
|-----------------------------------------+----------------------+----------------------+
| GPU  Name                 Persistence-M | Bus-Id        Disp.A | Volatile Uncorr. ECC |
| Fan  Temp   Perf          Pwr:Usage/Cap |         Memory-Usage | GPU-Util  Compute M. |
|                                         |                      |               MIG M. |
|=========================================+======================+======================|
|   0  NVIDIA GeForce RTX 3050 ...    Off | 00000000:01:00.0  On |                  N/A |
| N/A   47C    P5               5W /  60W |    509MiB /  4096MiB |      8%      Default |
|                                         |                      |                  N/A |
+-----------------------------------------+----------------------+----------------------+
                                                                                          
+---------------------------------------------------------------------------------------+
| Processes:                                                                            |
|  GPU   GI   CI        PID   Type   Process name                            GPU Memory |
|        ID   ID                                                             Usage      |
|=======================================================================================|
|    0   N/A  N/A      1623      G   /usr/lib/xorg/Xorg                          238MiB |
|    0   N/A  N/A      2917      G   /usr/bin/gnome-shell                         62MiB |
|    0   N/A  N/A      5596      G   /usr/lib/firefox/firefox                    136MiB |
|    0   N/A  N/A    111082      G   ...sion,SpareRendererForSitePerProcess       63MiB |
+---------------------------------------------------------------------------------------+

本地glxinfo | grep -i opengl输出：

OpenGL vendor string: NVIDIA Corporation
OpenGL renderer string: NVIDIA GeForce RTX 3050 Laptop GPU/PCIe/SSE2
OpenGL core profile version string: 4.6.0 NVIDIA 535.161.07
OpenGL core profile shading language version string: 4.60 NVIDIA
OpenGL core profile context flags: (none)
OpenGL core profile profile mask: core profile
OpenGL core profile extensions:
OpenGL version string: 4.6.0 NVIDIA 535.161.07
OpenGL shading language version string: 4.60 NVIDIA
OpenGL context flags: (none)
OpenGL profile mask: (none)
OpenGL extensions:
OpenGL ES profile version string: OpenGL ES 3.2 NVIDIA 535.161.07
OpenGL ES profile shading language version string: OpenGL ES GLSL ES 3.20
OpenGL ES profile extensions:

容器内glxinfo | grep -i opengl输出：

OpenGL vendor string: Mesa/X.org
OpenGL renderer string: llvmpipe (LLVM 12.0.0, 256 bits)
OpenGL core profile version string: 4.5 (Core Profile) Mesa 21.2.6
OpenGL core profile shading language version string: 4.50
OpenGL core profile context flags: (none)
OpenGL core profile profile mask: core profile
OpenGL core profile extensions:
OpenGL version string: 3.1 Mesa 21.2.6
OpenGL shading language version string: 1.40
OpenGL context flags: (none)
OpenGL extensions:
OpenGL ES profile version string: OpenGL ES 3.2 Mesa 21.2.6
OpenGL ES profile shading language version string: OpenGL ES GLSL ES 3.20
OpenGL ES profile extensions:

可正常运行的docker run命令：

sudo docker run -it --privileged --gpus all -e NVIDIA_DRIVER_CAPABILITIES=all  -e "DISPLAY=$DISPLAY" -v /tmp/.X11-unix:/tmp/.X11-unix -v /dev:/dev --name noetic_nvidia osrf/ros:noetic-desktop-full /bin/bash

解决方法

当前docker-compose配置存在两个问题：缺少NVIDIA_DRIVER_CAPABILITIES=all环境变量，且低版本docker-compose（1.28.0）对deploy.resources.reservations.devices的支持有限，需调整配置：

在environment中添加NVIDIA_DRIVER_CAPABILITIES=all，启用所有NVIDIA驱动能力（包括OpenGL渲染）
替换deploy.resources部分，改用runtime: nvidia（对应docker run --gpus all参数，低版本docker-compose中更可靠）

修改后的docker-compose.yml：

version: '3.7'

services:
  noetic:
    container_name: noetic
    build:
      context: ../
      dockerfile: .devcontainer/Dockerfile
    privileged: true
    network_mode: "host"
    runtime: nvidia  # 对应--gpus all参数
    command: /bin/bash
    volumes:
      - /dev:/dev
      - ../workspace:/workspace
      - /tmp/.X11-unix:/tmp/.X11-unix
    environment:
      - NVIDIA_VISIBLE_DEVICES=all
      - NVIDIA_DRIVER_CAPABILITIES=all  # 添加该环境变量
      - TZ=Asia/Shanghai
      - xpack.monitoring.enabled=false
      - xpack.watcher.enabled=false
      - DISPLAY=$DISPLAY
      - GDK_SCALE
      - GDK_DPI_SCALE
    deploy: {}  # 移除原GPU资源预留配置，如需CPU/内存限制可保留对应部分
    healthcheck:
      test: ["CMD", "curl", "-f", "http://localhost"]
      interval: 10m
      timeout: 20s
      retries: 3
    logging:
      driver: "json-file"
      options:
        max-size: "50M"
        max-file: "10"
    tty: true

修改后重新构建并启动容器：

docker-compose down
docker-compose build
docker-compose up -d

进入容器后再次执行glxinfo | grep -i opengl，即可看到NVIDIA GPU的渲染信息。

内容的提问来源于stack exchange，提问作者尹若尘

如何在docker-compose.yml中让容器内OpenGL调用NVIDIA GPU

问题背景 #

环境信息 #

文件夹结构 #

Dockerfile内容 #

Docker Compose版本 #

当前docker-compose.yml内容 #

问题现象 #

解决方法 #