Found this post, after a git pull update, Odoo 18.0 branch. Reverted changes to previous commit and the error is gone. I'm sure this is a recent bug to be identified yet.

I’m setting up a Digital Ocean VPS with 2GB RAM and 1 vCPU to run Odoo 18 Community. The server displays an error message “Real-time connection lost…” constantly on the bottom-right corner of the screen. I’ve changed the worker number in the Odoo Server configuration file and restarted the server, but the error persists. Interacting with the server indicates no issues. I can install the Odoo module or change settings. However, the message remains, suggesting a potential server connection problem which makes it impossible to know if there is a connection problem. Is this a bug?

PS: BTW, I have followed the tutorial video on this link: https://www.youtube.com/watch?v=Q1c5XpG85OE

I has same problem on odoo+nginx proxy mode configuration. any one can fix this?

I just fought with this exact issue for 2 straight days before I finally found the solution to my specific use case, so maybe this will help someone else.

I am running Odoo 18 Community, and its Postgres database, inside docker containers on Ubuntu. Both are created by a docker-compose.yml file. I am using Nginx as a reverse proxy installed on the main Ubuntu os, and Cloudflare for DNS and domain hosting.

The "accepted answer" for what to include in the Nginx config and the Odoo config are both a large part of what I needed, but the final needle in the haystack for my setup was that my docker-compose needed to also include port 8072 in it. I've also seen port 8071 indicated in places for something, so I added that too just in case. So my docker-compose is now:

version: '3'
services:
​odoo:
​ ​image: odoo:18.0
​ ​env_file: .env
​ ​depends_on:
​ ​ ​- postgres
​ ​ports:
​ ​ ​- "127.0.0.1:8069:8069"
​ ​ ​- "127.0.0.1:8072:8072"
​ ​ ​- "127.0.0.1:8071:8071"
​ ​volumes:
​ ​ ​- data:/var/lib/odoo
​ ​ ​- ./addons:/mnt/extra-addons
​ ​ ​- ./config:/etc/odoo
​ ​restart: unless-stopped
​postgres:
​ ​image: postgres:16
​ ​env_file: .env
​ ​volumes:
​ ​ ​- db:/var/lib/postgresql/data/pgdata
​ ​restart: unless-stopped

volumes:
​data:
​db:

Now that yellow box and message is gone :)

for all those struggling with nginx, just follow official documentation and posted solution
don't forget following block to enable connection upgrade in /websocket location

}
map $http_upgrade $connection_upgrade {
  default upgrade;
  ''      close;
}

in your ngnix conf add /websocket block

location /websocket {

    proxy_pass http://odoochat;

    proxy_set_header Upgrade $http_upgrade;

    proxy_set_header Connection $connection_upgrade;

    proxy_set_header X-Forwarded-Host $http_host;

    proxy_set_header X-Forwarded-For $proxy_add_x_forwarded_for;

    proxy_set_header X-Forwarded-Proto $scheme;

    proxy_set_header X-Real-IP $remote_addr;

    add_header Strict-Transport-Security "max-age=31536000; includeSubDomains";

#    proxy_cookie_flags session_id samesite=lax secure;  # requires nginx 1.19.8

  }

please do not forget to define these above your server block in ngninx

#odoo server

upstream odoo {

  server 127.0.0.1:8069;

}

upstream odoochat {

  server 127.0.0.1:8072;

}

map $http_upgrade $connection_upgrade {

  default upgrade;

  ''      close;

}

How would I do with with IIS on Windows?

I'm having the same issue. It is driving me crazy. I'm using Bitnami's Odoo package bitnami com/stack/odoo/virtual-machine on a VM Google Cloud instance. This package uses Apache as proxy. I thought it had to do with the SSL certificates, so I generated SSL for the subdomain I'm using but that didn't work either. The "Real-time error" is still there and the chat does not work in real time, neither calls & video calls. Not sure what else might be affected with this "Real-time error".

Below the 400 error related to the websocket not working:

2025-01-27 05:54:34,051 77881 INFO bitnami_odoo werkzeug: 201.152.145.175 - - [27/Jan/2025 05:54:34] "GET /websocket?version=18.0-3 HTTP/1.1" 400 - 1 0.001 0.002

2025-01-27 05:54:37,942 77879 INFO bitnami_odoo werkzeug: 201.152.145.175 - - [27/Jan/2025 05:54:37] "GET /websocket?version=18.0-3 HTTP/1.1" 400 - 1 0.001 0.003

2025-01-27 05:54:39,253 77879 INFO bitnami_odoo werkzeug: 201.152.145.175 - - [27/Jan/2025 05:54:39] "GET /websocket?version=18.0-3 HTTP/1.1" 400 - 1 0.001 0.002

2025-01-27 05:54:40,693 77879 INFO bitnami_odoo werkzeug: 201.152.145.175 - - [27/Jan/2025 05:54:40] "GET /websocket?version=18.0-3 HTTP/1.1" 400 - 1 0.001 0.002

I tried troubleshooting with ChatGPT and Gemini with no luck. Would be great to know if this is a bug, so I could stop trying to figure a workaround. Any advice or update would be greatly appreciated!

I using nginx as a reverse proxy with HTTPS for my Odoo18 setup.

In the odoo.conf file:

• When I set workers = (CPU*2)+1, the Nginx domain works fine, but accessing Odoo directly via http://{odoo_ip}:8069 throws a "Real-time connection lost" error.
• On the other hand, if I do not set the workers parameter, the Nginx domain throws an error, but accessing Odoo directly via http://{odoo_ip}:8069 works without issues.

Hi,
I think this is about longpolling port is set and gevent port is not set.

I am fixed it with.
In odoo.conf .

longpolling_port = False
gevent_port = 8072

you can also look at ( I cant share Link because of karma. )
deprecationwarning-the-longpolling-port-is-a-deprecated-alias-to-the-gevent-port-option-please-use-the-latter-214918

thanks Niyas ! but this is slightly different case in the YouTube video (man funny how people make it a 5 minutes videos these days! it's like one change in conf for number of worker processes :) 

What Fabio posted might be it, but kind of weird how would it be related

Colin were you able to find a fix for this? What does it mean anyways, is this connection to real-time chat, or something like this? 

PS tried Onur's suggestion but that didn't work for me..

Hi,
Can you check and see if this helps: Fix Connection Lost and Connection Restored Error Odoo

Thanks