连接GCP SQL实例时Nginx出现502 Bad Gateway问题求助

将Node.js应用部署在搭载Nginx的GCP VM实例上，无需数据库操作的接口可正常响应，但访问需要连接GCP Postgres SQL实例的API路由时，出现502 Bad Gateway错误。

环境配置信息 #

Nginx 配置 #

server {
     listen 80 default_server;
     listen [::]:80 default_server;

     root /var/www/html;

     # Add index.php to the list if you are using PHP
     index index.html index.htm index.nginx-debian.html;

     server_name _;

     location / {
             # First attempt to serve request as file, then
             # as directory, then fall back to displaying a 404.
            #try_files $uri $uri/ =404;

     proxy_pass http://localhost:5000;
     proxy_http_version 1.1;
     proxy_set_header Upgrade $http_upgrade;
     proxy_set_header Connection 'upgrade';
     proxy_set_header Host $host;
     proxy_cache_bypass $http_upgrade;
     }
}

PM2 应用日志 #

/root/.pm2/logs/edvive-out.log last 15 lines:
0|edvive   | 14/Dec/2022:09:50:22 +0000 "GET /api/v1/users/gcp-data"
0|edvive   | 14/Dec/2022:09:50:23 +0000 "GET /api/v1/users/gcp-data"
0|edvive   | user api hitted
0|edvive   | The server is running at port: 5000
0|edvive   | user api hitted
0|edvive   | The server is running at port: 5000
0|edvive   | 14/Dec/2022:09:51:01 +0000 "GET /api/v1/users/gcp-data"
0|edvive   | 14/Dec/2022:09:51:30 +0000 "GET /api/v1/users/gcp-data"
0|edvive   | user api hitted
0|edvive   | The server is running at port: 5000
0|edvive   | user api hitted
0|edvive   | The server is running at port: 5000
0|edvive   | user api hitted
0|edvive   | The server is running at port: 5000
0|edvive   | 14/Dec/2022:10:04:13 +0000 "GET /"

GCP Postgres 授权网络 #

Authorized networks
You can specify CIDR ranges to allow IP addresses in those ranges to access your instance. Learn more

production (35.***.**.***)
local (118.***.***.**)

Node.js 数据库连接配置 #

DATABASE_URL="postgresql://postgres:password@34.***.***.***:5432/postgres"

Nginx 错误日志 #

2022/12/14 10:03:20 [error] 5650#5650: *201 upstream prematurely closed connection while reading response header from upstream, client: 118.***.***.***, server: _, request: "GET /api/v1/users/gcp-data HTTP/1.1", upstream: "http://localhost:5000/api/v1/users/gcp-data", host: "35.***.**.***"
2022/12/14 10:03:20 [error] 5650#5650: *201 connect() failed (111: Connection refused) while connecting to upstream, client: 118.***.***.**, server: _, request: "GET /api/v1/users/gcp-data HTTP/1.1", upstream: "http://localhost/api/v1/users/gcp-data", host: "35.***.**.***"

1. 验证VM到Postgres的网络连通性 #

虽然VM IP已加入授权网络，仍需在VM内部直接测试连接：

• 执行telnet检查端口连通：

  telnet 34.***.***.*** 5432
  

• 或用psql直接测试数据库连接：

  psql postgresql://postgres:password@34.***.***.***:5432/postgres
  

如果连接失败：

• 确认GCP VM防火墙规则未拦截出站5432端口
• 检查Postgres授权网络中添加的是VM的公网IP（公网连接场景）或私网IP（VPC内网连接场景），避免混淆
• 若Postgres用私有IP，确保VM与Postgres在同一VPC，或配置了VPC peering

2. 查看Node.js应用的数据库错误日志 #

PM2的out.log仅记录请求到达，需查看应用错误日志定位数据库连接问题：

cat /root/.pm2/logs/edvive-error.log

若日志出现ECONNREFUSED、连接超时等报错：

• 核对DATABASE_URL中的Postgres IP、端口、密码（注意密码特殊字符需转义）
• 重新生成Prisma客户端：

  npx prisma generate
  

• 重启PM2应用确保环境变量生效：

  pm2 restart edvive
  

3. 修复Nginx配置潜在问题 #

从错误日志看存在上游地址缺失端口的异常，需确保Nginx配置生效：

• 检查配置语法：

  sudo nginx -t
  

• 重载配置：

  sudo nginx -s reload
  

同时建议添加超时配置，避免数据库查询超时导致502：

location / {
     proxy_pass http://localhost:5000;
     proxy_http_version 1.1;
     proxy_set_header Upgrade $http_upgrade;
     proxy_set_header Connection 'upgrade';
     proxy_set_header Host $host;
     proxy_cache_bypass $http_upgrade;
     # 添加超时配置
     proxy_connect_timeout 60s;
     proxy_send_timeout 60s;
     proxy_read_timeout 60s;
}

4. 验证应用本地访问 #

在VM内部直接调用API，确认问题是否出在应用本身：

curl http://localhost:5000/api/v1/users/gcp-data

若curl返回错误或超时，说明问题核心在应用与数据库的连接，而非Nginx反向代理，需重点排查应用数据库逻辑和网络连通性。

内容的提问来源于stack exchange，提问作者Md Firoz Mia