AI系列1-1：离线部署通义大模型及持续修正-RedHat+NVIDA GPU

原创

已于 2025-06-28 14:59:36 修改 · 972 阅读

17 ·

CC 4.0 BY-SA版权

文章标签：

#eureka #云原生

于 2025-06-25 20:23:52 首次发布

[题记]2025/6/21-6/22 两天辛苦的成果，有的是靠回忆写的，欢迎指正

1.环境基础

Redhat

查看redhat版本

cat /etc/redhat-release 7.9 Maipo

docker和docker compose安装

参考

https://zhuanlan.zhihu.com/p/14648196830

docker下载地址

https://download.docker.com/linux/static/stable/x86_64/

docker-compose下载地址

https://github.com/docker/compose/releases/tag/v2.37.1

安装过程

考虑到很多公司禁止使用知乎，再复制一下

（1）将docker-26.0.0.tgz压缩包与docker-compose-linux-x86_64上传到 /tmp 目录下

（2）解压docker-26.0.0.tgz压缩包并复制到/usr/bin/路径

tar -zxvf docker-*.tgz

sudo cp docker/* /usr/bin/

（3）对docker-compose-linux-x86_64文件重命名并添加可执行权限

mv docker-compose-* docker-compose

chmod +x docker-compose

（4）将docker-compose复制到/usr/bin并查看版本

sudo cp docker-compose /usr/bin/

docker-compose version

（5）创建docker.service文件

cd /etc/systemd/system/

touch docker.service

文件内容

[Unit]

Description=Docker Application Container Engine

Documentation=https://docs.docker.com

After=network-online.target firewalld.service

Wants=network-online.target


[Service]

Type=notify

# the default is not to use systemd for cgroups because the delegate issues still

# exists and systemd currently does not support the cgroup feature set required

# for containers run by docker

ExecStart=/usr/bin/dockerd

ExecReload=/bin/kill -s HUP $MAINPID

# Having non-zero Limit*s causes performance problems due to accounting overhead

# in the kernel. We recommend using cgroups to do container-local accounting.

LimitNOFILE=infinity

LimitNPRO