跳至主要內容

在服务器上部署进行上线

Cactus li2025年10月21日...大约 4 分钟Spring AISpring AIAgent

一、介绍

在 2 核 4GB 云服务器上完成 AI Agent 的整体部署，包含前端（HTML）、管理端（React）、后端（Java），以及 MySQL 与 PostgreSQL 向量数据库环境的搭建与配置。

二、部署过程

如图，为本次的部署过程；

Ai-Agent部署流程.drawio
Ai-Agent部署流程.drawio
  • 注意在云服务器上，需要开放端口 9000（Portainer）、8099（服务端）、3002（管理端）、8899（mysql管理端） 等你需要对外的端口。

三、工程结构

1. 服务端（含用户端）

image-20251201165434824
image-20251201165434824

构建整个工程，分为1步 ~ 4步，前面3步是构建和上传镜像，第4步，是部署项目。教程里的第4步会在云服务器操作。

2. 管理端

image-20251105154044926
image-20251105154044926

管理端部署前需调整 api.ts 中的 API 地址，修改完成后再进行构建。 若企业环境中已接入统一的 API 网关，则无需修改 IP，系统会通过网关域名自动路由至对应服务。完成配置后，依次执行以下命令进行镜像的构建与推送：

./build.sh
./push.sh

四、部署

1. 镜像构建

1.1 服务端

docker build -t cactuslixf/ai-agent-station-study-app:1.0 -f ./Dockerfile .

[+] Building 0.3s (10/10) FINISHED                                                                                               
 => [internal] load build definition from Dockerfile                                                                             
 => => transferring dockerfile: 1.29kB                                                                                           
 => WARN: MaintainerDeprecated: Maintainer instruction is deprecated in favor of using label (line 5)                           
 => [internal] load metadata for docker.io/library/openjdk:17-ea-17-jdk-slim-buster                                             
 => [internal] load .dockerignore                                                                                               
 => => transferring context: 2B                                                                                                 
 => [1/5] FROM docker.io/library/openjdk:17-ea-17-jdk-slim-buster                                                               
 => [internal] load build context                                                                                               
 => => transferring context: 87B                                                                                                 
 => CACHED [2/5] RUN ln -snf /usr/share/zoneinfo/PRC /etc/localtime && echo PRC > /etc/timezone                                 
 => CACHED [3/5] RUN echo "deb http://archive.debian.org/debian buster main" > /etc/apt/sources.list &&     echo "deb http://archive.debian.org/debian-se  0.0s
 => CACHED [4/5] RUN node -v && npm -v                                                                                           
 => CACHED [5/5] ADD target/ai-agent-station-study-app.jar /ai-agent-station-study-app.jar                                       
 => exporting to image                                                                                                           
 => => exporting layers                                                                                                         
 => => writing image sha256:7a8f62a8ddb574a5e5456275a998f6ca418c819445086c118b643f73da968b12                                     
 => => naming to docker.io/cactuslixf/ai-agent-station-study-app:1.0
docker push cactuslixf/ai-agent-station-study-app:1.0

The push refers to repository [docker.io/cactuslixf/ai-agent-station-study-app]
2c3da3a39327: Pushed
12656d7bae56: Pushed
91bc063eac36: Pushed
7494202df645: Pushed
06379473cd2c: Mounted from library/openjdk
2651e73b1464: Mounted from library/openjdk
7e718b9c0c8c: Mounted from library/openjdk
1.0: digest: sha256:6d643f8d11dee4e2bfe913827493b03418687c262a05f0eb880aef2a9ac44639 size: 1791
  • 执行了构建脚本，并上传镜像到 Docker Hub镜像库。

1.2 管理端

docker build  -t cactuslixf/ai-agent-station-admin-app:1.0 -f ./Dockerfile .

[+] Building 109.2s (22/22) FINISHED                                                                                             
 => [internal] load build definition from Dockerfile
 => => transferring dockerfile: 2.90kB                                                                                           
 => WARN: FromAsCasing: 'as' and 'FROM' keywords' casing do not match (line 2)                                                   
 => [internal] load metadata for docker.io/library/nginx:1.25.1                                                                 
 => [internal] load metadata for docker.io/library/node:20-alpine                                                               
 => [internal] load .dockerignore                                                                                               
 => => transferring context: 2B                                                                                                 
 => [internal] load build context                                                                                               
 => => transferring context: 4.12MB                                                                                             
 => [base 1/1] FROM docker.io/library/node:20-alpine                                                                             
 => [runner 1/5] FROM docker.io/library/nginx:1.25.1                                                                             
 => CACHED [runner 2/5] WORKDIR /usr/share/nginx/html                                                                           
 => CACHED [runner 3/5] RUN rm -rf ./*                                                                                           
 => CACHED [deps 1/6] RUN apk add --no-cache libc6-compat                                                                       
 => CACHED [deps 2/6] WORKDIR /app                                                                                               
 => CACHED [deps 3/6] COPY package.json ./                                                                                       
 => CACHED [deps 4/6] COPY pnpm-lock.yaml* yarn.lock* package-lock.json* ./                                                     
 => [deps 5/6] RUN npm install                                                                                                   
 => [deps 6/6] RUN npm install @rspack/[email protected] @rspack/[email protected] --save-dev                                               
 => CACHED [builder 1/4] WORKDIR /app                                                                                           
 => CACHED [builder 2/4] COPY --from=deps /app/node_modules ./node_modules                                                       
 => [builder 3/4] COPY . .                                                                                                       
 => [builder 4/4] RUN npx rsbuild build                                                                                         
 => [runner 4/5] COPY --from=builder /app/dist .                                                                                 
 => [runner 5/5] RUN echo 'server {' > /etc/nginx/conf.d/default.conf &&     echo '    listen 3002;' >> /etc/nginx/conf.d/default.conf &&     echo '    server_name local  0.5s
 => exporting to image                                                                                                           
 => => exporting layers                                                                                                         
 => => writing image sha256:e1c2345f35f3d343985493d7b73acc9d3d3141c3e65926714a1ca216db637135                                     
 => => naming to docker.io/cactuslixf/ai-agent-station-admin-app:1.0                                                             
 1 warning found (use docker --debug to expand):
 - FromAsCasing: 'as' and 'FROM' keywords' casing do not match (line 2)
'cactuslixf/ai-agent-station-admin-app:1.0 Dockerfile: Dockerfile' has been deployed successfully.
docker push cactuslixf/ai-agent-station-admin-app:1.0

The push refers to repository [docker.io/cactuslixf/ai-agent-station-admin-app]
b51e08b6b4a1: Pushed
1a196142f820: Pushed
046cf5bb583e: Pushed
5f70bf18a086: Mounted from jiangjuhong/dujiaoka
922d16116201: Mounted from library/nginx
abc3beec4b30: Mounted from library/nginx
c88d3a8ff009: Mounted from library/nginx
8aedfcd777c7: Mounted from library/nginx
4deafab383fa: Mounted from library/nginx
24ee1d7d6a62: Mounted from library/nginx
c6e34807c2d5: Mounted from library/nginx
1.0: digest: sha256:ced4a7960f7349706b6ed2025efda1b084caebb8cbde5c37b39e6a84aed120fd size: 2609
  • 执行了构建脚本，并上传镜像到 Docker Hub镜像库。

2. 服务部署

2.1 部署脚本

# /usr/local/bin/docker-compose -f /docs/dev-ops/environment/environment-docker-compose-app.yml up -d
version: '3.8'
# docker-compose -f docker-compose-app.yml up -d
services:
  # 部署前端项目
  ai-agent-station-front-app:
    image: cactuslixf/ai-agent-station-front-app
    container_name: ai-agent-station-front-app
    restart: always
    ports:
      - '80:80'
    volumes:
      - ./nginx/html:/usr/share/nginx/html
    privileged: true
    networks:
      - my-network

  # 管理后台
  ai-agent-station-admin-app:
    image: cactuslixf/ai-agent-station-admin-app:1.0
    container_name: ai-agent-station-admin-app
    ports:
      - "3002:3002"
    restart: unless-stopped
    networks:
      - my-network
    healthcheck:
      test: [ "CMD", "curl", "-f", "http://localhost:3002" ]
      interval: 30s
      timeout: 10s
      retries: 3
      start_period: 40s
    labels:
      - "traefik.enable=true"
      - "traefik.http.routers.ai-agent-front.rule=Host(`localhost`)"
      - "traefik.http.services.ai-agent-front.loadbalancer.server.port=3002"

  # 进入后端命令 docker exec -it ai-agent-station-study-app /bin/bash
  ai-agent-station-study-app:
    image: cactuslixf/ai-agent-station-study-app:1.1
    container_name: ai-agent-station-study-app
    restart: on-failure
    ports:
      - "8099:8099"
    environment:
      - TZ=PRC
      - SERVER_PORT=8091
      - SPRING_DATASOURCE_MYSQL_USERNAME=root
      - SPRING_DATASOURCE_MYSQL_PASSWORD=123456
      - SPRING_DATASOURCE_MYSQL_URL=jdbc:mysql://mysql:3306/ai-agent-station-study?serverTimezone=UTC&characterEncoding=utf8&autoReconnect=true&serverTimezone=Asia/Shanghai
      - SPRING_DATASOURCE_MYSQL_DRIVER_CLASS_NAME=com.mysql.cj.jdbc.Driver
      - SPRING_DATASOURCE_PGVECTOR_DRIVER_CLASS_NAME=org.postgresql.Driver
      - SPRING_DATASOURCE_PGVECTOR_USERNAME=postgres
      - SPRING_DATASOURCE_PGVECTOR_PASSWORD=postgres
      - SPRING_DATASOURCE_PGVECTOR_URL=jdbc:postgresql://vector_db:5432/ai-rag-knowledge?connectTimeout=120
      - SPRING_AI_OPENAI_BASE_URL=https://apis.itedus.cn
      - SPRING_AI_OPENAI_API_KEY=sk-Vb7vnKNtGFAsKGhz6765DfE3F1974908Bb6650D7169d5aD7
      - SPRING_AI_OPENAI_AGENT_AUTO-CONFIG_ENABLED=true
    volumes:
      - ./log:/data/log
    logging:
      driver: "json-file"
      options:
        max-size: "10m"
        max-file: "3"
    networks:
      - my-network

networks:
  my-network:
    driver: bridge
  • ai-agent-station-front-app，是通过 nginx 部署的前端 html 页面。
  • ai-agent-station-admin-app，是部署的管理后台，注意这个镜像地址要更换为你的 cactuslixf/ai-agent-station-admin-app:1.0
  • ai-agent-station-study-app，是部署的服务端，这里要注意，SPRING_AI_OPENAI_BASE_URL、SPRING_AI_OPENAI_API_KEY 要更换为你的。

2.2 上传脚本

image-20251118162702315
image-20251118162702315
  • 可以通过 https://termius.com/ 工具（免费）的 sftp 能力，把脚本上传到服务器上。

2.3 执行脚本

2.3.1 环境安装
docker compose -f docker-compose-environment.yml up -d
image-20251118174132776
image-20251118174132776
2.3.2 后端部署 + toc用户端 + 管理端
docker compose  -f docker-compose-app.yml up -d
image-20251118174650596
image-20251118174650596
  • 部署项目，包括；服务端（study）、toc用户端（front）、管理端（admin）
默认页脚
Copyright © 2018-2025 Cactus li