使用代理时 Nginx 502 Bad Gateway 错误

Question

我有一个 Angular 构建和一个 Laravel 后端，提供在一台服务器上运行的 API。我已经在 nginx 中配置了它们，前端有一个后端服务器的代理。

后端在 url（例如占位符）http://api.example.com 上运行，前端在 http://example.com 上运行

前端配置：

server {
    listen       80;
    server_name  example.com;

    location /api {
        proxy_pass http://api.example.com;
        proxy_http_version 1.1;
        proxy_set_header Upgrade $http_upgrade;
        proxy_set_header Connection "upgrade";
        proxy_set_header Host $host;
    }

    location / {
        root  /var/www/angular/em-frontend/dist;
        index  index.html index.htm;
        try_files $uri $uri/ /index.html$is_args$args;
    }
}

后端配置：

server {
        listen 80;
        server_name api.example.com;

        root /var/www/angular/em-backend/public;

        index index.php index.html index.htm;

        location / {
                # First attempt to serve request as file, then
                # as directory, then fall back to displaying a 404.
                try_files $uri $uri/ /index.php?$query_string;
                # Uncomment to enable naxsi on this location
                # include /etc/nginx/naxsi.rules
        }
        location ~ \.php$ {
                try_files $uri =404;
                fastcgi_split_path_info ^(.+\.php)(/.+)$;
                fastcgi_pass unix:/var/run/php/php7.0-fpm.sock;
                fastcgi_index index.php;
                fastcgi_param SCRIPT_FILENAME $document_root$fastcgi_script_name;
                include fastcgi_params;
        }
}

现在，当我从前端进行任何 api 调用时，我从 nginx 收到 502 Bad Gateway 错误。

来自 nginx 错误日志：

2017/12/09 23:30:40 [alert] 5932#5932: 768 worker_connections are not enough
2017/12/09 23:30:40 [error] 5932#5932: *770 recv() failed (104: Connection reset by peer) while reading response header from upstream, client: IP_MASKED, server: example.com, request: "GET /api/endpoint HTTP/1.1", upstream: "http://IP_ADDRESS:80/api/endpoint", host: "example.com", referrer: "http://example.com/dashboard"

知道如何解决这个问题吗？

Answer 1

你必须像这个例子一样在 location 块中使用 proxy-pass：

upstream myproject {
   server ip1 ;
   server ip2 ;
   server ip3 ;

} 
location / { 
   proxy_pass      http://myproject;
}

Answer 2

我相信您的问题是主机名配置创建了一个递归循环，其中一个请求被代理回前端，很快就会耗尽所有工作人员。您将通过对前端的单个请求在访问日志中生成许多条目来识别这一点。

我能够使用您提供的配置快速重现该错误。下面是一个修改后的版本，它消除了在后端服务器上提供 2 个不同静态文件的配置，以说明所需的最低配置。如果可行，您可以重新添加 cgi_pass 配置。

#set api domain to use alternate port, could also just tack onto proxy_pass.
upstream api.example.com {
    server localhost:8081;
}

#frontend listening on port 8080
server {
    listen       8080;
    server_name  example.com;

    location /api {
        proxy_pass http://api.example.com;
        proxy_http_version 1.1;
        proxy_set_header Upgrade $http_upgrade;
        proxy_set_header Connection "upgrade";
        proxy_set_header Host $host;
    }

    location / {
        root  /usr/local/var/www;
        index  index.html index.htm;
        try_files $uri $uri/ /index.html$is_args$args;
    }
}

#backend listening on 8081
server {
        listen 8081;
        server_name api.example.com;

        index index.php index.html index.htm;

        location / {  # will match any url not ending in .php
              root  /usr/local/var/www;
              try_files $uri $uri/ /index.html;
        }
        location ~ \.php { #successfully responds to http://example.com:8080/api/*.php
              root  /usr/local/var/www;
              try_files $uri $uri/ /service.html;
        }
}

Answer 3

我遇到了类似的问题，解决了：

events {
    worker_connections 1024;
}

events 块与http 块处于同一级别。