Hi everyone,
I’m facing a frustrating issue with my self-hosted n8n instance.
Problem
-
The editor constantly shows workflows as “Offline” (Connection lost / No network connection).
-
Old workflows continue to work fine (executions run, Telegram bot responds quickly).
-
New workflows cannot be properly saved/published — they stay in a broken state.
-
The backend itself is healthy (/health returns 200), service is live, and executions work.
Environment
-
n8n version: 2.19.5 (tried 2.20.9 before — same issue)
-
Hosting: Render.com (free + paid instance tried)
-
Database: Supabase Postgres (free tier)
-
Docker image: n8nio/n8n:2.19.5
What I already tried
-
Correct WEBHOOK_URL, N8N_EDITOR_BASE_URL, N8N_PROXY_HOPS=1
-
N8N_PUSH_BACKEND=sse and WebSocket
-
N8N_ENDPOINT_HEALTH=health
-
Different n8n versions (2.19.5 ↔ 2.20.9)
-
N8N_RUNNERS_ENABLED=true + internal mode
-
Pruning old executions + reducing data size in workflows
-
Hard refresh, different browsers
Logs
I see a lot of these:
-
Slow database query
-
Custom data value over 512 characters long. Truncating…
-
Task runner errors: Failed to connect to n8n task broker at 127.0.0.1:5679 + Unexpected server response: 403
-
Database connection timed out / Connection terminated
Old projects work without issues, only newly created or recently edited workflows are affected.
Has anyone experienced similar problems on Render + Supabase recently?
Any working configuration for n8n 2.20+ on Render?
Thanks in advance!