MCP Server Trigger in last version doesn´t work correctly

[Claudi]

Working in 1.94.1 version in docker, when I aggregate a MCP Server trigger the URL appears correctly like: https://mydomain/mcp-test/decaabe7-cb67-4806-8761-a05fc14e942a/sse.

But when I upgrade my n8n to the latest version (1.100.1) the url has the following appearence:
https://mydomain/mcp-test/d7988ab5-6b4e-423a-8094-7690b8d74931

so without the /sse and then it doesn´t work.

[cutecatcode]

Hi there!
It works fine with my self-hosted n8n (Ofc with docker and latest version), and I can’t find any error, instead, it just running correctly.
Perhaps you forgot to Execute the MCP Server Trigger workflow when you testing it?

[Claudi]

This is my docker-compose file

services:
n8n:
image: docker.n8n.io/n8nio/n8n:latest # Última versión disponible
container_name: n8n
restart: unless-stopped
ports:
- “5678:5678” # Puerto público directo
environment:
# Configuración HTTPS
- N8N_HOST=mydomain.com
- N8N_PORT=5678
- N8N_PROTOCOL=https
- WEBHOOK_URL=https://mydomain.com

  # Configuración SSL con certificados externos
  - N8N_SSL_KEY=/certs/privkey.pem
  - N8N_SSL_CERT=/certs/fullchain.pem
  
  # Configuración de seguridad
  - N8N_SECURE_COOKIE=true
  - N8N_ENFORCE_SETTINGS_FILE_PERMISSIONS=true
  
  # Configuración de runners y proxy
  - N8N_RUNNERS_ENABLED=true
  - N8N_PROXY_HOPS=0  # Sin proxy
  
  # Configuración regional y técnica
  - GENERIC_TIMEZONE=Europe/Madrid
  - N8N_PUSH_BACKEND=websocket
  - N8N_PUSH_BACKEND_WS=true
  
  # Configuración adicional para HTTPS directo
  - N8N_LISTEN_ADDRESS=0.0.0.0
  
volumes:
  - ./n8n_data:/home/node/.n8n
  - ./n8n_export:/home/node/workflows_export
  - ./certs:/certs:ro  # Certificados SSL externos en modo solo lectura

[cutecatcode]

there is nothing wrong with that, can you give me screenshots of your MCP Server and Client tool?

[Claudi]

It is something strange. I think I have an issue with my public DNS because I can´t access now to n8n main page from another computer, just only locally on my server where I have deployed my docker. And although now nothing works because my https public doesn´t work, now I can see the URL correcly with /sse at the end

I have to fix this first, but yesterday when all was working fine I couldn´t see it… I have to wait until Monday for fixing the DNS issue and I will let you know

Thanks for your time and I will inform you on Monday

[Claudi]

Hi, Finally It worked. I see now the sse but when I create my client MCP and I write the URL the node doesn´t show me the mcp servers. Checking it in another deploy with npm and ngrok works correctly. Any idea?

[cutecatcode]

Do you use authentication for the MCP Server triggers? If yes, you need to turn it off

[Claudi]

No, I don´t use it