Nginx configuration - Node finishes but UI does not update causing a loop

[vadim_antemattr]

Currently experiencing issues where node finishes, but UI in n8n does not update, see linked video below:

raw_media_file-...

This is happening across different nodes and canvases. Jon mentioned in Discord that this is likely an issues with poorly configured Nginx on our server.

I did some extensive detective work to try and SSH tunnel into the railway server. but Railway abstracts most infrastructure stuff and does not offer a way to SSH into the server.

[n8n]

It looks like your topic is missing some important information. Could you provide the following if applicable.

• n8n version:
• Database (default: SQLite):
• n8n EXECUTIONS_PROCESS setting (default: own, main):
• Running n8n via (Docker, npm, n8n cloud, desktop app):
• Operating system:

[vadim_antemattr]

Here is the additional info:

Version 1.64.3
Database: Postgres + Redis
n8n executions: queue
Running n8n via docker image on Railway with workers
Mac Studio Sonoma 14.4

[Jon]

Hey @vadim_antemattr,

Welcome to the community

Did you check if you are using a reverse proxy at all? Quickly looking at it you are using SSE for the backend which should help so I wonder if maybe something is cached in the browser or there is some setting in Render causing an issue.

When running the workflow if you open the browser dev console do you get any errors?

[vadim_antemattr]

@Jon Will look at console later today but yesterday went through my deep maintenance routine and reinstalled Chrome fresh after purging and cleaning my machine with CleanMyMac and in new chrome the same issues are happening.

I did some research into how Railway manages their deployments, and currently they do not offer a way to SSH into anything, their CLI does not permit that. Then all of their NGINX stuff is abstracted away and not accessible in settings or dashboard.

By the way, we had two other completely separate servers running on Railway and have never had this problem before on there. The ENV settings have been replicated across all servers and all had just minor changes.

[Jon]

Hey @vadim_antemattr,

Nginx was just an example they may use something else, the important thing is making sure that websockets are not being blocked if they are being used.

Did you check the browser dev console for any errors? I would expect to maybe see something under the console or network section that could be useful.

[vadim_antemattr]

@Jon So changing the nginx.conf and modifying the dockerfile to import just the modified below config did not help unfortunately, same issues and even worse with this config settings, perhaps I didn’t quite get it right:

worker_processes auto;
events {
worker_connections 1024;
}
http {
include mime.types;
default_type application/octet-stream;

sendfile        on;
keepalive_timeout  65;

server {
    listen       80;
    server_name  primary-production-aa44.up.railway.app api.antemattr.com;

    location / {
        proxy_pass http://primary-production-ba56.up.railway.app;
        proxy_set_header Host $host;
        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_buffering off;
    }
}

}

the new docker file I pushed just had this.

FROM --platform=linux/amd64 n8nio/n8n

COPY nginx.conf /etc/nginx/nginx.conf

[vadim_antemattr]

@Jon In fact these changes and the custom n8n image made things worse, now when runingthings manually in canvas, after I run 5 or so nodes, all previous data drops completely.

This has been extremely frustrating as the workflow has come to a crawl in n8n.

Any additional troubleshooting steps that you can provide?

[vadim_antemattr]

@Jon please let me know some feedback when you have a moment as this is still not resolved