Help for owners
My n8n webhooks show the wrong address or do not trigger
Short answer
Behind a reverse proxy, n8n builds webhook addresses from its internal host and port, which the internet cannot reach. n8n's documentation says to set the public webhook address, tell n8n how many proxies are in front, and forward the standard forwarded headers. Also check that the workflow is published, because the test address works differently.
Does this sound like your situation?
- The webhook address in n8n shows localhost or port 5678.
- The service that sends me data says the address is unreachable.
- The test button works but real calls do nothing.
- It worked in my office but not from outside.
- Calls arrive but I cannot see them in the editor.
Why does this happen?
n8n builds the address from its own host and port
The docs say that without a public address setting n8n builds the webhook URL from its protocol, host and port. Behind a reverse proxy it listens on an internal port that the public cannot reach, so the shown address is wrong.
Confirmed by docs.n8n.io
The proxy setup is incomplete
The docs ask for the public webhook address to be set, the number of proxies in front of n8n to be set, and the X-Forwarded-For, X-Forwarded-Host and X-Forwarded-Proto headers to be forwarded on the last proxy.
Confirmed by docs.n8n.io
Test and production addresses are different
n8n registers the test address while you listen for a test event and the workflow is not active. The production address is registered when you publish the workflow, and calls to it do not show data in the editor; you see them in the Executions tab.
Confirmed by docs.n8n.io
The sending service gives up on slow or failing answers
The webhook node can answer immediately, when the last node finishes, or through a Respond to Webhook node. Waiting until the end of a long workflow can exceed what the sender allows. Payloads are also limited to 16MB unless a self-hosted setting changes it.
Confirmed by docs.n8n.io
What can I do about it today?
- Check the workflow is publishedA workflow that is only saved will not answer on its production address.
- Open the webhook address from a phone on mobile dataNot on your office network. A message from n8n, even an error, means the route is open; a time-out means it is not.
- Look at the Executions tabCalls to the production address show there, not in the editor.
- Send your developer the address and the sender's errorThe exact address shown in n8n, the one the service tried, and the error text or status code.
- Do not open extra ports on the server yourselfExposing n8n directly can make it less safe. Leave changes on the proxy to the person who manages it.
When should I bring someone in?
- The address still shows an internal host after you were told it was fixed.
- Payments, orders or leads depend on the webhook.
- You are unsure who manages the proxy or the certificate.
How do I handle it?
- Follow one call from outside to n8nSender, DNS, certificate, proxy, n8n: I test each hop to see where it stops.
- Set the public address and proxy countThe settings n8n documents are set to match your real domain and the number of proxies in front.
- Check the forwarded headers on the last proxyThe proxy configuration is read and corrected so n8n sees the original host, protocol and client address.
- Register production, test separatelyI confirm the sender uses the production address of a published workflow, and use the test address only while building.
- Answer fast, process laterFor slow work, the webhook is set to respond first and the long steps run afterwards.
The price depends on what I find, so I quote in writing after a short discovery call. Tell me what is happening and I reply within 24 hours.
What do owners ask about this?
Which setting changes the webhook address?
n8n's page describes a public webhook address variable, and notes the older name is deprecated. I check the page for your version before changing it.
Why does the test URL work but not the real one?
The two are registered differently. The production one works only after the workflow is published, and its calls appear in Executions.
Do I need HTTPS?
Most services that send webhooks expect a secure address. The setup page does not cover certificates, so that part is checked on your proxy.
Sources
How was this page made?
I drafted this page with AI assistance from the official documents listed under Sources. Found a mistake? Tell me.
For your developer: the technical fixes
502 Bad Gatewaynginx 502 Bad Gateway in front of a Node.js app504 Gateway Time-outnginx gives up waiting for the app and returns 504WebSocket closes with code 1006 after roughly 60 seconds of no messagesWebSockets behind nginx drop when the connection is idle400 Bad Requestnginx 400: a plain HTTP request reached the HTTPS port