Dify - 整合Ollama + Xinference私有化部署Dify平台（02 ）通过Docker安装Dify

总体方案

• ollama 和 Xinference 管理模型，需要充分利用GPU， 故这里不选择使用Docker安装。

• Dify 使用Docker安装

服务器

幕僚云按需租几台服务器

选择合适的镜像：机器学习框架镜像

开启学术加速代理

目前机器学习框架镜像默认集成学术加速，需要手动开启学术加速代理

请在终端中使用以下命令开启学术加速。

source /etc/network_turbo

备注：如果network_turbo文件不存在，北京一区、北京二区、北京三区等使用此命令设置相关代理

export http_proxy=http://10.132.19.35:7890

export https_proxy=http://10.132.19.35:7890

步骤预览

cd dify
cd docker
cp .env.example .env
docker compose up -d

1. 下载源码

(py38) root@vmInstancecmvl766w:~# df -h
Filesystem                         Size  Used Avail Use% Mounted on
udev                                56G     0   56G   0% /dev
tmpfs                               12G  9.3M   12G   1% /run
/dev/mapper/ubuntu--vg-ubuntu--lv   48G   18G   28G  38% /
tmpfs                               56G     0   56G   0% /dev/shm
tmpfs                              5.0M     0  5.0M   0% /run/lock
tmpfs                               56G     0   56G   0% /sys/fs/cgroup
/dev/sda2                          2.0G  378M  1.5G  21% /boot
/dev/sdb                            49G   24K   47G   1% /data
/dev/loop1                          64M   64M     0 100% /snap/core20/2501
/dev/loop2                          92M   92M     0 100% /snap/lxd/24061
/dev/loop0                          64M   64M     0 100% /snap/core20/2434
/dev/loop3                          92M   92M     0 100% /snap/lxd/29619
/dev/loop4                          45M   45M     0 100% /snap/snapd/23258
/dev/loop5                          45M   45M     0 100% /snap/snapd/23771
tmpfs                               12G     0   12G   0% /run/user/0
(py38) root@vmInstancecmvl766w:~#
(py38) root@vmInstancecmvl766w:~# cd /data/
(py38) root@vmInstancecmvl766w:/data#
(py38) root@vmInstancecmvl766w:/data# git clone  https://github.com/langgenius/dify.git
Cloning into 'dify'...
remote: Enumerating objects: 154717, done.
remote: Counting objects: 100% (878/878), done.
remote: Compressing objects: 100% (500/500), done.
remote: Total 154717 (delta 695), reused 378 (delta 378), pack-reused 153839 (from 5)
Receiving objects: 100% (154717/154717), 88.33 MiB | 14.53 MiB/s, done.
Resolving deltas: 100% (111349/111349), done.
(py38) root@vmInstancecmvl766w:/data#

2. 修改Dify端口号

查找 EXPOSE_NGINX_PORT ,修改为自己原因设置的端口号

(py38) root@vmInstancecmvl766w:/data/dify/docker# grep EXPOSE_NGINX_PORT  .env
EXPOSE_NGINX_PORT=8890

3. 使用docker启动Dify

docker compose up -d

修改： 进入/etc/docker/daemon.json文件 ，然后在里面加入下面的配置

{
  "registry-mirrors": ["https://docker.registry.cyou",
"https://docker-cf.registry.cyou",
"https://dockercf.jsdelivr.fyi",
"https://docker.jsdelivr.fyi",
"https://dockertest.jsdelivr.fyi",
"https://mirror.aliyuncs.com",
"https://dockerproxy.com",
"https://mirror.baidubce.com",
"https://docker.m.daocloud.io",
"https://docker.nju.edu.cn",
"https://docker.mirrors.sjtug.sjtu.edu.cn",
"https://docker.mirrors.ustc.edu.cn",
"https://mirror.iscas.ac.cn",
"https://docker.rainbond.cc"]
}
 

如果上面的源不行了，重新再换一个源，比如如下：

"https://docker.unsee.tech",
"https://dockerpull.org",
"https://docker.1panel.live",
"https://dockerhub.icu",
"https://docker.m.daocloud.io",
"https://docker.nju.edu.cn",
"https://registry.docker-cn.com",
"https://docker.mirrors.ustc.edu.cn",
"https://hub-mirror.c.163.com",
"https://mirror.baidubce.com",
"https://5tqw56kt.mirror.aliyuncs.com",
"https://docker.hpcloud.cloud",
"http://mirrors.ustc.edu.cn",
"https://docker.chenby.cn",
"https://docker.ckyl.me",
"http://mirror.azure.cn",
"https://hub.rat.dev"

然后在终端重新启动一下docker

systemctl daemon-reload

systemctl restart docker

重新启动

根据网速不同，等待片刻，即可

docker ps  

4. 本地访问Dify

因为用的是云平台，所以需要根据映射的端口来访问Dify的服务 ，根据映射

访问IP:24264