WS issue after TLS termination

[wey-gu]

Describe the bug (must be provided)

When studio is behind HTTPS/WSS, it'll got WebSocket ERROR.

Your Environments (must be provided)

• Studio-version: 3.7.0

How To Reproduce(must be provided)

Steps to reproduce the behavior:

With this docker compose yaml and nginx.conf, we could easily reproduced the issue.

version: '3.4'
services:
  web:
    image: vesoft/nebula-graph-studio:v3.7.0
    environment:
      USER: root
    ports:
      - 7001
    networks:
      - nebula-net
  nginx:
    image: nginx:1.17.6
    ports:
      - 7001:443
    volumes:
      - ./nginx.conf:/etc/nginx/nginx.conf
      - ./cert:/cert
    networks:
      - nebula-net

networks:
  nebula-net:
    external: true

nginx.conf:

events {
}

http {
    server {
        listen 443 ssl;
        server_name nebula-demo.siwei.io;
        ssl_certificate /cert/fullchain.pem;
        ssl_certificate_key /cert/privkey.pem;
        location / {
            proxy_pass http://web:7001;
        }

        location /nebula_ws {
            proxy_pass http://web:7001;
            proxy_http_version 1.1;
            proxy_set_header Upgrade $http_upgrade;
            proxy_set_header Connection "upgrade";
        }
    }
}

Expected behavior

working

Additional context

Screen Captured as follow:

Screen.Recording.2023-08-07.at.18.02.54.mp4

[wey-gu]

cc @hetao92 :-D

[sbocahu]

we reproduce the same issue with nginx-ingress controller with tls termination in front of a nebule-studio helm-deployed instance on kubernetes

any workaround or fix you know of ?

[salahaz]

@wey-gu any updates on this issue ?

[wey-gu]

After checking with @huaxiabuluo to know it's already fixed in 3.8.0 release, and I verified to confirm it's fixed :)

[wey-gu]

3.8.0 over wss in container

3.8.0.regression.mov

[huaxiabuluo]

@salahaz @sbocahu
Under the HTTPS protocol, the WebSocket service link prefix needs to be changed from ws:// to wss://, this bug has been fixed in version 3.8, please upgrade to the latest version.

[salahaz]

Thank you very much @wey-gu and @huaxiabuluo for your support!!! It is resolved.