How to Fix the Most Common n8n Errors

n8n is powerful and forgiving. When it breaks, it usually breaks the same handful of ways. Here's how to fix each.

Workflow doesn't trigger

Symptom: Webhook URL returns 200 but workflow doesn't execute.

Causes & fixes: 1. Workflow is "inactive" — toggle the switch top-right 2. You're testing on the test URL not production — check active URL in webhook node 3. Webhook path collision — change webhook ID in node settings

Node fails silently

Symptom: Workflow completes but data isn't where expected.

Causes & fixes: 1. Expression returns empty — add Set node before to log input 2. Pin sample data and re-run to inspect node-by-node 3. Add "On Error → Continue" temporarily to see all downstream behavior

OAuth keeps disconnecting

Symptom: Google/Slack/HubSpot credentials disconnect every 24 hours.

Causes & fixes: 1. Token refresh not configured — most OAuth nodes need a refresh token, check credentials 2. Provider revoked due to inactivity — re-authenticate 3. Self-hosted n8n missing N8N_HOST env var — OAuth callbacks need correct callback URL

Memory leak on self-hosted instance

Symptom: n8n container OOMs after 3-7 days.

Causes & fixes: 1. EXECUTIONS_DATA_PRUNE=true + EXECUTIONS_DATA_MAX_AGE=168 env vars (auto-prune week-old data) 2. Move execution data from SQLite to Postgres (DB_TYPE=postgresdb) 3. Set N8N_PAYLOAD_SIZE_MAX=16 (MB) to reject huge webhooks

"Cannot read property X of undefined" in expression

Symptom: Red node, error mentions undefined property.

Fix: Add ?? fallback in expression: {{ $json.data?.user?.email ?? "unknown" }}. n8n uses standard JS optional chaining.

Workflow runs but no email/notification

Fix: Check provider quotas. Slack rate-limits at 1 msg/sec. SendGrid has daily caps on free tier. Gmail has 500 sends/day via Apps Script.

Hire an n8n developer →