解决docker: Error response from daemon: could not select device driver ““ with capabilities: [[gpu]].

25-02-16 06:20

12046

blog.csdn.net

id="article_content" class="article_content clearfix" style="height: 2000px; overflow: hidden;"> id="content_views" class="htmledit_views">

这个错误

could not select device driver "" with capabilities: [[gpu]] class="hljs-button signin active" data-title="登录复制" data-report-click="{"spm":"1001.2101.3001.4334"}" onclick="hljs.signin(event)">

通常是由于 Docker 没有正确识别到 GPU，或者 NVIDIA Docker 配置不正确。

一、检查主机 GPU 和 NVIDIA 驱动是否正常工作

执行

nvidia-smi class="hljs-button signin active" data-title="登录复制" data-report-click="{"spm":"1001.2101.3001.4334"}" onclick="hljs.signin(event)">

确认主机的 GPU 运行环境是否正常

如果命令输出错误，需要先确保 NVIDIA 驱动正确安装

二、检查 NVIDIA 容器工具包是否安装

执行

dpkg -l | grep nvidia-container-toolkit class="hljs-button signin active" data-title="登录复制" data-report-click="{"spm":"1001.2101.3001.4334"}" onclick="hljs.signin(event)">

确认 NVIDIA 容器工具包是否已正确安装

如果没有任何信息出现，则使用以下命令安装

 class="hljs-ln-numbers"> class="hljs-ln-line hljs-ln-n" data-line-number="1">

class="hljs-ln-code"> class="hljs-ln-line">distribution=$(. /etc/os-release;echo $ID$VERSION_ID) \

 class="hljs-ln-numbers"> class="hljs-ln-line hljs-ln-n" data-line-number="2"> class="hljs-ln-code"> class="hljs-ln-line">    && curl -s -L https://nvidia.github.io/nvidia-docker/gpgkey | sudo apt-key add - \

 class="hljs-ln-numbers"> class="hljs-ln-line hljs-ln-n" data-line-number="3"> class="hljs-ln-code"> class="hljs-ln-line">    && curl -s -L https://nvidia.github.io/nvidia-docker/$distribution/nvidia-docker.list | sudo tee /etc/apt/sources.list.d/nvidia-docker.list

 class="hljs-ln-numbers"> class="hljs-ln-line hljs-ln-n" data-line-number="4"> class="hljs-ln-code"> class="hljs-ln-line">

 class="hljs-ln-numbers"> class="hljs-ln-line hljs-ln-n" data-line-number="5"> class="hljs-ln-code"> class="hljs-ln-line">sudo apt update

 class="hljs-ln-numbers"> class="hljs-ln-line hljs-ln-n" data-line-number="6"> class="hljs-ln-code"> class="hljs-ln-line">sudo apt install -y nvidia-container-toolkit

 class="hljs-ln-numbers"> class="hljs-ln-line hljs-ln-n" data-line-number="7"> class="hljs-ln-code"> class="hljs-ln-line">sudo systemctl restart docker

class="hljs-button signin active" data-title="登录复制" data-report-click="{"spm":"1001.2101.3001.4334"}" onclick="hljs.signin(event)">

三、Docker运行时使用 NVIDIA配置

sudo vim /etc/docker/daemon.json class="hljs-button signin active" data-title="登录复制" data-report-click="{"spm":"1001.2101.3001.4334"}" onclick="hljs.signin(event)">

添加以下内容（如果已有其他配置，保留其他配置）：

 class="hljs-ln-numbers"> class="hljs-ln-line hljs-ln-n" data-line-number="1"> class="hljs-ln-code"> class="hljs-ln-line">{
 class="hljs-ln-numbers"> class="hljs-ln-line hljs-ln-n" data-line-number="2"> class="hljs-ln-code"> class="hljs-ln-line">  "default-runtime": "nvidia",
 class="hljs-ln-numbers"> class="hljs-ln-line hljs-ln-n" data-line-number="3"> class="hljs-ln-code"> class="hljs-ln-line">  "runtimes": {
 class="hljs-ln-numbers"> class="hljs-ln-line hljs-ln-n" data-line-number="4"> class="hljs-ln-code"> class="hljs-ln-line">    "nvidia": {
 class="hljs-ln-numbers"> class="hljs-ln-line hljs-ln-n" data-line-number="5"> class="hljs-ln-code"> class="hljs-ln-line">      "path": "nvidia-container-runtime",
 class="hljs-ln-numbers"> class="hljs-ln-line hljs-ln-n" data-line-number="6"> class="hljs-ln-code"> class="hljs-ln-line">      "runtimeArgs": []
 class="hljs-ln-numbers"> class="hljs-ln-line hljs-ln-n" data-line-number="7"> class="hljs-ln-code"> class="hljs-ln-line">    }
 class="hljs-ln-numbers"> class="hljs-ln-line hljs-ln-n" data-line-number="8"> class="hljs-ln-code"> class="hljs-ln-line">  }
 class="hljs-ln-numbers"> class="hljs-ln-line hljs-ln-n" data-line-number="9"> class="hljs-ln-code"> class="hljs-ln-line">} class="hljs-button signin active" data-title="登录复制" data-report-click="{"spm":"1001.2101.3001.4334"}" onclick="hljs.signin(event)">

编辑完长这样

在vim界面，输入:wq退出编辑状态

四、保存后重启 Docker

sudo systemctl restart docker class="hljs-button signin active" data-title="登录复制" data-report-click="{"spm":"1001.2101.3001.4334"}" onclick="hljs.signin(event)">

删除之前的容器，重新启动容器，及可调用GPU

注：本文转载自blog.csdn.net的ZhouDevin的文章"https://blog.csdn.net/ZhouDevin/article/details/145335593"。版权归原作者所有，此博客不拥有其著作权，亦不承担相应法律责任。如有侵权，请联系我们删除。

复制链接

发表评论

/ 登录

评论记录：

未查询到任何数据！