N8N Community on Ubuntu - could not connect to MCP server

Questions
1

Hey all, I’m having the same issue everybody else is having, except I’ve disabled gzip for the /SSE path and I’m still unable to connect to my MCP server trigger.

If I test the MCP server URL, I’m getting a 404.

Any ideas on how to fix this on an Ubuntu server? It seems everybody else is using Docker.

2

Specify which URL you are using and the method used to test it.

Indicate if there are any n8n or server error logs.

Mention how you are running n8n on Ubuntu (commands, environment, services).

Share how you disabled gzip (exact configuration).

1 Like
3

No error logs that I can see. I can create a standalone MCP Server trigger and try to test, but I get a 404 not found. The URL for the MCP trigger is https://n8n.redacted/mcp-test/ghl-mcp/sse

I’m running n8n on a VPS by itself on Ubuntu via the web interface.

In the nginx config, I have this to disable gzip:

location ~* /sse$ {
	gzip off;
}

location / {
	proxy_pass http://localhost:5678;
	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_cache off;
	proxy_buffering off;
}

Thanks for any advice you can provide!

1 Like
4

404 when accessing /sse: This indicates that the endpoint does not exist from the server’s perspective, which could be due to: The MCP trigger is not properly registered or activated within N8N, and the full path /mcp-test/ghl-mcp/sse is not being correctly interpreted by NGINX or N8N.

NGINX and GZIP configuration: The rule location ~* /sse$ disables gzip only if the URL ends exactly in /sse. However, if the URL is /mcp-test/ghl-mcp/sse, this rule does not apply.

Change the rule to:

location ~* /sse {
gzip off;
}

N8N Trigger Verification: Make sure the trigger is activated. Verify that it was saved and deployed correctly. Try a shortened version of the URL to isolate the problem.

Then:
Restart NGINX after applying the changes.
sudo nginx -t && sudo systemctl reload nginx
Check the N8N logs
Run local tests on port 5678 to verify the endpoint is responding without going through NGINX

1 Like