使用自定义域名在Google Compute Engine上进行Websocket连接

本文介绍了使用自定义域名在Google Compute Engine上进行Websocket连接的处理方法，对大家解决问题具有一定的参考价值，需要的朋友们下面随着小编来一起学习吧！ 问题描述

我有一个websocket服务器(node.js)，可以在localhost和以前的heroku部署上正常运行.我现在正在迁移到Google计算引擎，并遇到了一些问题.

I have a websocket server (node.js) which runs fine on localhost and on a previous heroku deployment. I am now migrating to google compute engine and running into some issues.

websocket握手失败，返回301错误.正如此答案中指出的，这可能是由于请求通过了不支持Websocket连接的前端服务器而导致的.可以通过直接定位 ws://my_external_gce_ip 来解决.我想知道是否可以更新一些负载平衡配置，以便可以使用自定义域名访问后端.

The websocket handshake is failing returning a 301 error. As pointed out in this answer this can be due to the request going through front end servers that do not support a websocket connection and can be worked around by targeting ws://my_external_gce_ip directly. I am wondering if there is some load balancing configuration I can update so that I can address my backend using the custom domain name.

虽然我了解问题所在，但在我看来，应该在dns查找之后将域解析为外部ip，这样我才不会真正理解约束.

While I understand the problem, it seems to me that the domain should be resolved to the external ip after a dns lookup so I don't understand the constraint really.

很抱歉，如果这很明显.我是GCE的新手，并且整日都在搜寻Google，以获取此信息.我将在下面粘贴我的代码以及NGINX配置，但是我认为这两个都不是特别有用，因为所有这些都可以使用IP很好地寻址

Sorry if this is very obvious. I'm new to GCE and have been googling all day trying to get this. I will paste my code below as well as the NGINX config but I don't think either are particularly helpful as all works fine addressing using the IP

index.js:

/* requirements */ var bodyParser = require("body-parser"); const WebSocket = require("ws"); const http = require("http"); const express = require("express"); const port = process.env.PORT || 3000; /* server definition and config */ const app = express(); app.use(bodyParser.json()); const server = http.createServer(app); /* web socket stuff */ const webSocketServer = new WebSocket.Server({ server, }); webSocketServer.on("connection", (webSocket) => { console.log("board trying to connect..."); webSocket.on("message", (data) => { webSocketServer.clients.forEach((client) => { if (client === webSocket && client.readyState === WebSocket.OPEN) { client.send("[SERVER MESSAGE]: You are connected to the server :)"); } }); }); }); /* activate server */ server.listen(port, () => { console.log(`Server is now running on port ${port}\n`); });

nginx配置

server { listen 80 default_server; listen [::]:80 default_server; root /var/www/html; index index.html index.htm index.nginx-debian.html; server_name _; location \ { # we're actually going to proxy all requests to # a Nodejs backend proxy_pass localhost:8080; 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; # I added this baby in proxy_set_header X-Forwarded-For $proxy_add_x_forwarded_for; } } server { root /var/www/html; index index.html index.htm index.nginx-debian.html; server_name my_domain; # managed by Certbot location / { # we're actually going to proxy all requests to # a Nodejs backend proxy_pass localhost:8080; proxy_http_version 1.1; proxy_set_header Upgrade $http_upgrade; proxy_set_header Connection 'upgrade'; proxy_set_header Host $host; proxy_cache_bypass $http_upgrade; } listen [::]:443 ssl ipv6only=on; # managed by Certbot listen 443 ssl; # managed by Certbot ssl_certificate /etc/letsencrypt/live/my_domain/fullchain.pem; # managed by Certbot ssl_certificate_key /etc/letsencrypt/live/my_domain/privkey.pem; # managed by Certbot include /etc/letsencrypt/options-ssl-nginx.conf; # managed by Certbot ssl_dhparam /etc/letsencrypt/ssl-dhparams.pem; # managed by Certbot } server { if ($host = my_domain) { return 301 $host$request_uri; } # managed by Certbot listen 80 ; listen [::]:80 ; server_name my_domain; return 404; # managed by Certbot }

非常感谢，如果这是一个菜鸟问题，我们深表歉意.负载均衡方面的新手

Thanks very much in advance and sorry if this is a noob question. Total noob when it comes to load balancing

推荐答案

已解决.经过更多的谷歌搜索后，我发现此主题的人大多数都面临相同的问题使用apache服务器或弹性beantalk，因此不使用nginx.

Solved. After a lot more googling I found this thread of people facing the same problem most either using apache servers or elastic beanstalk so not using nginx.

似乎很多人都通过Websocket来工作".使用socket.io，但它们并没有真正的双工连接，因为它可以回退到长时间轮询.

It seems that a lot of people get websockets to "work" using socket.io but they don't really have a duplex connection as it is falling back to long polling.

就我而言，答案很简单，我没有在我的nginx(facepalm)中包含服务器名称，并且我可能忘记了包含标头.现在，https转发看起来像这样(并且可以使用我的域进行寻址)

In my case the answer was simple, I didn't include the server name in my nginx (facepalm) and I may have forgot to include a header. The https forwarding now looks like this (and addressing using my domain works)

location / { # we're actually going to proxy all requests to # a Nodejs backend proxy_pass localhost:8080; 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; proxy_set_header X-Real-IP $remote_addr; proxy_set_header X-Forwarded-For $proxy_add_x_forwarded_for; proxy_set_header X-Forwarded-Proto $scheme; proxy_read_timeout 90; }

别忘了在更新后重新启动nginx

Don't forget to restart your nginx after you update

sudo systemctl restart nginx