Hello Railway Community,

I am running a self-hosted n8n instance in Queue Mode (Primary + Worker + Postgres + Redis) on Railway.

Current status:

• Deployments of Primary and Worker are successful

• All workflow executions run normally in the background (triggers fire, workflows complete successfully)

• However, the n8n Editor UI permanently shows "Connection Lost" in the top right corner

• I cannot save or manually execute any workflows in the UI

• The Railway Architecture view only shows a black/blank box with a GitHub icon and never loads the service cards (happening for >24 hours) (see screenshot)

• I cannot access the service settings or detailed configuration through the Railway dashboard
  
  
  Important note: The problem started after usage of MCP Server Trigger + external API/MCP calls from IDE. Since then the UI WebSocket/SSE connection fails consistently, while background executions continue to work.
  
  Does anyone know

  • Whether there is a problem with WebSocket/SSE proxying on our Primary service

  • Why the Architecture dashboard is stuck showing only a black box

  • Any internal errors or resource limits affecting the Primary service

  Thank you for your help!