Docker部署Nginx反向代理Django应用服务器端502故障排查

问题描述 #

通过Docker部署Nginx反向代理的Django应用，本地运行正常，但部署到服务器后网页显示502 Bad Gateway错误，Nginx日志如下：

2023/01/13 07:55:56 [error] 28#28: *1 connect() failed (111: Connection refused) while connecting
to upstream, client: xx.xxx.xx.xxx, server: , request: "GET /favicon.ico HTTP/1.1", upstream:
"http://xxx.xx.x.x:8000/favicon.ico", host: "xx.xxx.xxx.xxx", referrer: "http://xxx.xxx.xxx.xxx/"

相关配置文件 #

Nginx Dockerfile #

FROM nginx:1.23.3-alpine
RUN rm /etc/nginx/conf.d/default.conf
COPY ./default.conf /etc/nginx/conf.d/default.conf

Nginx default.conf #

upstream apii {
    server api:8000;
}

server {
    client_max_body_size 20M;
    listen 80;

    location / {
        proxy_pass http://apii;
        proxy_set_header X-Forwarded-For $proxy_add_x_forwarded_for;
        proxy_set_header Host $host;
        proxy_redirect off;
    }

    # location = /favicon.ico { access_log off; log_not_found off; }

    location /api {
        proxy_pass http://apii;
        proxy_set_header X-Forwarded-For $proxy_add_x_forwarded_for;
        proxy_set_header Host $host;
        proxy_redirect off;
    }

    location /realestateadmin {
        proxy_pass http://apii;
        proxy_set_header X-Forwarded-For $proxy_add_x_forwarded_for;
        proxy_set_header Host $host;
        proxy_redirect off;

    }

    location /staticfiles/ {
        alias /app/staticfiles/;
    }

    location /mediafiles/ {
        alias /app/mediafiles/;
    }

}

docker-compose.yml #

version: "3.9"

services:
    api:
        build:
            context: .
            dockerfile: ./docker/local/django/Dockerfile
        command: /start
        volumes:
            - .:/app
            - static_volume:/app/staticfiles
            - media_volume:/app/mediafiles
        # ports:
        #     - "8000:8000"
        expose:
            - 8000
        env_file:
            - .env
        depends_on:
            - postgres-db
        networks:
            - real-estate

    postgres-db:
      image: postgres:12.0-alpine
      ports:
          - "5432:5432"
      volumes:
          - postgres_data:/var/lib/postgresql/data/
      environment:
          - POSTGRES_USER=${DB_USER}
          - POSTGRES_PASSWORD=${DB_PASSWORD}
          - POSTGRES_DB=${DB_NAME}
      networks:
          - real-estate
    
    nginx:
        restart: always
        depends_on:
            - api
        volumes:
            - static_volume:/app/staticfiles
            - media_volume:/app/mediafiles
        build: 
            context: ./docker/local/nginx
            dockerfile: Dockerfile
        ports:
          - "80:80"
        networks:
            - real-estate

networks:
  real-estate:
    driver: bridge

volumes:
  postgres_data:
  static_volume:
  media_volume:

排查思路 #

• 确认Django服务状态

  • 查看api容器启动日志：docker-compose logs api，检查是否有启动失败、数据库连接超时或端口绑定异常
  • 进入api容器内部，验证服务监听状态：docker-compose exec api netstat -tulpn | grep 8000，或用curl localhost:8000测试本地访问是否正常

• 检查Docker网络连通性

  • 在nginx容器内测试访问api服务：docker-compose exec nginx curl http://api:8000，如果无法访问，确认两个容器是否都加入了real-estate网络
  • 查看容器网络归属：docker inspect api | grep Networks 和 docker inspect nginx | grep Networks，确保网络一致

• 验证Django绑定地址

  • 检查Django启动脚本（/start），确保服务绑定的是0.0.0.0:8000，而非127.0.0.1:8000。如果绑定localhost，容器外部（包括nginx）无法访问该端口

• 调整服务启动顺序

  • depends_on仅保证postgres-db先启动，但不保证数据库完全就绪。可在api的启动脚本中加入数据库健康检查，等待postgres可连接后再启动Django
  • 手动重启api容器：docker-compose restart api，观察是否能正常连接数据库并启动服务

• 校验Nginx配置

  • 检查nginx容器内的配置文件：docker-compose exec nginx cat /etc/nginx/conf.d/default.conf，确认upstream指向api:8000
  • 测试配置语法正确性：docker-compose exec nginx nginx -t，确保无语法错误

• 排查服务器网络限制

  • 确认服务器80端口对外开放，容器间的8000端口无需对外暴露，但要保证同一Docker网络内的容器可互相访问
  • 检查服务器iptables规则或云服务商安全组，是否限制了容器间的网络通信

内容的提问来源于stack exchange，提问作者Abdul Fertah