Contents
- The Short Answer
- The Two URLs, and Why There Are Two
- Cause 1: You Called the Test URL and Nothing Was Listening
- Cause 2: You Called the Production URL of a Workflow That Is Not Published
- Cause 3: The HTTP Method Does Not Match
- Cause 4: The Path Is Not the One n8n Registered
- Cause 5: The Publish Did Not Fully Succeed
- Cause 6: Self-Hosted Configuration
- How to Stop It Coming Back
- Sources
- FAQ
The Short Answer
"The requested webhook is not registered" is a 404 that n8n returns when a request arrives for a webhook path it is not currently listening on. The full response looks like this:
{
"code": 404,
"message": "The requested webhook \"POST my-path\" is not registered.",
"hint": "..."
}
Read the hint first. It tells you which of the two usual causes you have:
- If the hint says to click the Execute workflow button on the canvas (older versions say Test workflow), you called the Test URL and nobody was listening. Start a test in the editor and send the request again, or switch to the Production URL.
- If the hint says the workflow must be active for a production URL to run successfully, you called the Production URL of a workflow that is not live. Publish the workflow.
If neither applies, the cause is the HTTP method, the path, a publish that did not fully succeed, or the way a self-hosted instance is set up. Each is covered below.
The Two URLs, and Why There Are Two
n8n generates two Webhook URLs for each Webhook node. The n8n Webhook documentation summarises the difference:
| Test URL | Production URL | |
|---|---|---|
| Default path | /webhook-test/<path> | /webhook/<path> |
| How to make it listen | Select Listen for test event, then trigger an event from the source | Publish the workflow |
| How long it listens | 120 seconds | Until the workflow is unpublished |
| Data shown in the editor | Yes | No. Look in the executions list |
The split exists so you can build and debug against real requests without your half-finished workflow being live. It also means there are two separate registrations, and a request to either one fails when its registration is not in place.
A note on wording. Current n8n uses Publish and Unpublish. Earlier versions used an Active toggle in the top right of the editor, and the error hint in the source still says "active". They refer to the same state: the workflow's triggers are registered and running.
The error text and both hints come from one file in the n8n source, webhook-not-found.error.ts, which is worth a look if you want the exact strings.
Cause 1: You Called the Test URL and Nothing Was Listening
The hint for this case reads: "Click the 'Execute workflow' button on the canvas, then try again. (In test mode, the webhook only works for one call after you click this button)".
Two limits are packed into that. The test webhook is only registered after you start listening in the editor, and per the docs it stays active for 120 seconds. And per the hint, it accepts one call. A second request to the same Test URL, or a first request that arrives after the window closes, gets the 404.
How to confirm it: the URL you are calling contains /webhook-test/.
The fix while building:
- Open the Webhook node and select Listen for test event, or run the workflow from the canvas.
- Send the request within the listening window.
- For each further test, start listening again first.
The n8n docs give curl commands you can use to send a test request from a terminal. Replace the URL with your own:
curl --request POST https://your-n8n.url/webhook-test/path --data 'key=value'
The fix for anything that should keep working: never give a Test URL to an external service. Stripe, a form tool, another team's system: all of them need the Production URL. This is the single most common version of the problem. The Test URL was pasted into the other service during setup, it worked once while the editor was open, and it has returned 404 ever since.
Cause 2: You Called the Production URL of a Workflow That Is Not Published
The hint for this case reads: "The workflow must be active for a production URL to run successfully. You can activate the workflow using the toggle in the top-right of the editor. Note that unlike test URL calls, production URL calls aren't shown on the canvas (only in the executions list)".
The n8n publishing documentation states what publishing turns on: webhook and form triggers use their production URLs, schedules run, and events from connected apps trigger the workflow. Until then, the production path is not registered.
How to confirm it: the URL contains /webhook/ and not /webhook-test/. On the Workflows page, a published workflow shows an indicator on its card. If yours does not, it is not live.
The fix: open the workflow and select Publish in the canvas header. The docs note the shortcut is Shift + P. Then send the request again.
Two things trip people up after this. First, n8n saves your edits automatically as a draft, but production executions use the published version, not your latest edits. If you change the Webhook node's path or method, the live webhook keeps the old values until you publish again. Second, as the hint says, production calls do not appear on the canvas. A workflow can be receiving requests perfectly well while the editor shows nothing. Check the executions list before deciding it is not working.
Cause 3: The HTTP Method Does Not Match
When the path is registered but not for the method you used, n8n returns a different and more helpful message:
This webhook is not registered for GET requests. Did you mean to make a POST request?
The classic way to hit this is pasting the webhook URL into a browser address bar. A browser sends a GET. If the Webhook node is set to POST, you get this message, and the webhook is fine.
How to confirm it: compare the HTTP Method in the Webhook node with the method your caller sends. In curl, that is the value after --request.
The fix: change one side to match the other. If the caller is a third-party service, its documentation says which method it sends, and the node should follow. If you need the same path to accept more than one method, the docs give the steps:
- Open the Webhook node's Settings.
- Turn on Allow Multiple HTTP Methods.
- Return to Parameters. By default the node now accepts GET and POST. Add other methods in the HTTP Methods field.
The node then has an output for each method, so you can handle them differently.
Cause 4: The Path Is Not the One n8n Registered
The message quotes the method and path n8n looked up. If that path is not character for character what the Webhook node shows, the request is going somewhere that does not exist.
How to confirm it: open the Webhook node, select the Production URL, copy it, and compare it with the URL the caller is using. Common differences:
- The URL was typed or assembled by hand rather than copied.
- The path in the node was edited after the URL was given to the other service, and the workflow was published again with the new path.
/webhook-test/was changed to/webhook/by hand and something else in the path was changed with it.
The fix: always copy the URL from the node and paste it into the calling service. If you need a stable, readable address, set the Path field in the Webhook node deliberately, publish, and treat that path as a contract that does not change.
Related: n8n permits only one webhook for each path and HTTP method combination. If you try to publish a second workflow using a path and method that are already taken, n8n tells you they are in use. The docs give two ways out: unpublish the workflow with the conflicting webhook, or change the path or method on one of them.
Cause 5: The Publish Did Not Fully Succeed
You published, the URL is exact, the method matches, and the 404 is still there. Check whether the triggers actually registered. Publishing is asynchronous: n8n registers triggers in the background and the Publish button shows Publishing until it has a result. The publishing docs describe two outcomes other than success:
- Published, partial: n8n published the new version, but at least one trigger failed to activate, for example because a connected app was unavailable. The workflow keeps running the triggers that did activate.
- Failed to publish: none of the workflow's triggers activated, so the workflow is not running.
How to confirm it: look at the Publish button in the canvas header. In the partial state, hovering over it shows which triggers failed.
The fix: publish again to retry. In both states the docs say the published version stays set, so a second publish retries the failed triggers. If a trigger keeps failing, the reason is in that trigger's own configuration or credential, and fixing that comes first. The docs also describe a state where the button shows an error because recent changes contain problems that must be fixed before you can publish. Production executions always use the currently published version, so until you fix and publish, what is live is not what you see on the canvas.
Cause 6: Self-Hosted Configuration
If you run n8n yourself and the checks above all pass, look at the deployment.
The URL in the node is wrong for the outside world. n8n builds the webhook URL from N8N_PROTOCOL, N8N_HOST and N8N_PORT. The reverse proxy documentation explains that this does not work behind a proxy, because n8n runs internally on port 5678 while the proxy exposes it on 443. If the Webhook node shows a URL with localhost or port 5678, external services cannot be given a correct address from it. Set the public URL explicitly:
export N8N_WEBHOOK_URL=https://n8n.example.com/
export N8N_PROXY_HOPS=1
The docs note that N8N_WEBHOOK_URL is available from n8n 2.30.0 and replaces WEBHOOK_URL, which is deprecated from 2.35.0. On earlier versions, use WEBHOOK_URL. The same page lists the headers the last proxy must pass on: X-Forwarded-For, X-Forwarded-Host and X-Forwarded-Proto.
The proxy does not send webhook paths to n8n. If the 404 you receive is not the JSON shown at the top of this page, it did not come from n8n. Make sure the proxy forwards /webhook/ and /webhook-test/ to the instance. Those are the default paths. They can be changed with N8N_ENDPOINT_WEBHOOK and N8N_ENDPOINT_WEBHOOK_TEST, so check whether yours have been.
Two deployments share one Redis. n8n's deployment documentation warns that if you point two n8n deployments at the same Redis host without a unique key prefix for each, they share the same queue and leader-election channels, and activating a workflow on one instance can produce webhook not registered errors on the other.
How to confirm any of these: call the production URL with curl from the server itself, against n8n's internal address and port. If that works and the public URL does not, the problem is between the internet and n8n, not in the workflow.
Hosting details like these are a real part of the running cost of self-hosted n8n. Our guide to the true cost of self-hosting n8n covers what you take on, and the glossary entry on self-hosting explains the trade-off in general.
How to Stop It Coming Back
Give external services the Production URL only. Make it a rule. The Test URL is for you, in the editor, for a couple of minutes at a time.
Publish after every change to the Webhook node. Path, method and authentication changes do nothing to the live webhook until you publish again.
Check the executions list, not the canvas. Production calls are not shown in the editor. Before concluding a webhook is broken, look for the execution.
Keep a one-line health check. Save a curl command for each important webhook next to the workflow, in a sticky note. When someone reports a problem, run it. A 200 tells you the fault is on the sender's side. A 404 with n8n's JSON tells you to work through this page.
Respond quickly on n8n Cloud. A separate failure worth knowing about: the docs say n8n Cloud sits behind Cloudflare, and if a webhook does not respond within 100 seconds the request fails with a 524 status code. For long-running work, respond straight away and do the work afterwards.
Webhooks are how most production n8n workflows start, including chat and messaging agents. The n8n AI Agents course builds several webhook-triggered agents end to end, and the Telegram AI agent tutorial is a free example of a trigger that registers a webhook for you. For when to use a webhook rather than a schedule, see polling vs webhook, and for the platform overall, our n8n review.
Sources
Checked against n8n's documentation, source code and community forum on 17 November 2026.
- n8n docs: Webhook common issues - Test URL versus Production URL, multiple methods, one webhook per path and method, curl examples, Cloud timeout.
- n8n docs: Webhook workflow development - Listen for test event and the 120 second window.
- n8n docs: Save and publish workflows - what publishing enables, and the partial and failed states.
- n8n docs: Configure webhook URLs with reverse proxy -
N8N_WEBHOOK_URL,N8N_PROXY_HOPS, default paths. - n8n source: webhook-not-found.error.ts - the exact message and both hints.
- n8n community: Webhook production URL only working in test - the test path versus production path mix-up in practice.
FAQ
- What does "The requested webhook is not registered" mean in n8n?
- n8n received an HTTP request for a webhook path it is not currently listening on, and answered with a 404. Either the Test URL was called when no test was listening, the Production URL was called for a workflow that is not published, or the path or method does not match what the Webhook node registered.
- What is the difference between the n8n Test URL and Production URL?
- The Test URL, on the /webhook-test/ path, listens only after you select Listen for test event in the editor, for 120 seconds, and shows the incoming data in the editor. The Production URL, on the /webhook/ path, listens for as long as the workflow is published and does not show data in the editor.
- Why does my webhook work once and then return 404?
- You are calling the Test URL. According to the hint n8n returns, in test mode the webhook only works for one call after you start the test. Use the Production URL and publish the workflow for anything that needs to keep working.
- My workflow is published and I still get the 404. What now?
- Copy the Production URL from the Webhook node and compare it exactly with the one being called, check the HTTP method, then look at the Publish button for a partial or failed publish and publish again. On self-hosted instances, check that your reverse proxy forwards the webhook paths to n8n.
- What does "This webhook is not registered for GET requests" mean?
- The path exists but is registered for a different HTTP method. It often appears when a webhook URL is opened in a browser, which sends a GET, while the Webhook node expects POST. Match the method, or turn on Allow Multiple HTTP Methods in the node's Settings.
- Why can I not see production webhook calls on the canvas?
- That is by design. Production URL calls are not shown in the editor. They appear in the executions list. Only Test URL calls display their data on the canvas.
- Which environment variable sets the public webhook URL on self-hosted n8n?
- N8N_WEBHOOK_URL, available from n8n 2.30.0. It replaces WEBHOOK_URL, which the docs say is deprecated from 2.35.0. Behind a reverse proxy, also set N8N_PROXY_HOPS to 1 and forward the X-Forwarded-For, X-Forwarded-Host and X-Forwarded-Proto headers.
If this resonated
Module 02 is the long version of this post.
Eval set, refusal pattern, kill switch. Build it in a weekend.
