n8n reference

n8n errors, explained one by one

11documented messages
5categories

Last verified

n8n reports errors at the node level. When a node fails, the execution stops (unless the node is set to continue on error), the node turns red on the canvas, and the error panel shows a short message plus details. Two kinds of node errors dominate. API errors come from the external service: n8n maps common HTTP statuses to friendly messages such as "Bad request - please check your parameters" (400), "Forbidden - perhaps check your credentials" (403), "The resource you are requesting could not be found" (404) and "The service is receiving too many requests from you" (429), and the raw response from the service sits in the error details. Operation errors come from n8n or your own logic, such as a Code node throwing "Cannot read properties of undefined".

Some errors are not tied to a single run. Webhook URLs can be "not registered" because the test URL only listens briefly and the production URL only works while the workflow is active. Activation can fail because a trigger cannot register with its service. A workflow with unresolved node issues refuses to run at all. And on self-hosted instances, memory limits can kill the process, which the editor reports as "Execution stopped at this node".

Each entry below quotes the message as n8n shows it, explains what it means and what is not wrong, and lists fixes drawn from n8n's documentation and the issues people report in the community forum. Where behavior differs between n8n Cloud and self-hosted, the entry says so.

Official reference: n8n Docs - Error handling

How to read a n8n error

In an execution, click the failed node. Error workflows and notifications summarize it as "Problem in node" plus the node's name:

Problem in node 'HTTP Request'
Forbidden - perhaps check your credentials        <- n8n's message for the HTTP status

Error details
  httpCode: 403                                   <- status from the service
  rawErrorMessage: 403 - {"message":"..."}        <- the service's own explanation

NodeApiError means the external service returned an error; NodeOperationError means the node itself (or your configuration or code) failed. Read the raw message before changing anything.

All 11 messages

Jump to: Errors from external services (5) · Webhooks (1) · Activation and validation (2) · Code node (1) · Memory and crashes (2)

Errors from external services

MessageHTTPType
429 - The service is receiving too many requests from youThe service is receiving too many requests from you429Retryable
Bad request - please check your parameters400Configuration
Forbidden - perhaps check your credentials403Configuration
The resource you are requesting could not be found404Configuration
Connection refused (ECONNREFUSED)connect ECONNREFUSED 127.0.0.1:8080—Configuration

Webhooks

MessageHTTPType
The requested webhook "..." is not registeredThe requested webhook "POST my-path" is not registered.404Configuration

Activation and validation

MessageHTTPType
Workflow could not be activatedWorkflow could not be activated:—Configuration
The workflow has issues and cannot be executedThe workflow has issues and cannot be executed for that reason. Please fix them first.—Configuration

Code node

MessageHTTPType
Cannot read properties of undefined (reading '...') [line N]Cannot read properties of undefined (reading 'length') [line 2]—Configuration

Memory and crashes

MessageHTTPType
Execution stopped at this node (n8n may have run out of memory)Execution stopped at this node (n8n may have run out of memory while executing it)—Configuration
FATAL ERROR: JavaScript heap out of memoryFATAL ERROR: Reached heap limit Allocation failed - JavaScript heap out of memory—Configuration

Definitions

n8n guides, limits and tools

Other error directories

Frequently asked questions

What does "Problem in node" mean in n8n?

It is the header n8n uses when a node fails, followed by the node's name. It is not an error by itself: the line after it is the actual message, such as a 403 or a Code node TypeError. Open the execution, click the named node, and read the error details, including the raw response from the service.

What is the difference between NodeApiError and NodeOperationError?

NodeApiError wraps an error response from an external API, so the fix is usually in your request, credentials or the service's limits. NodeOperationError is raised by the node itself, for example invalid configuration, bad input data or a failed operation in n8n. The error details show which one occurred.

Why does my n8n webhook work in testing but not in production?

Test and production URLs are different. The test URL contains /webhook-test/ and only listens after you start a test, for a single call. The production URL contains /webhook/ and only works while the workflow is active. External services must use the production URL, and the workflow must be activated.

How do I retry failed nodes in n8n?

In a node's Settings, enable Retry On Fail and set Max Tries and Wait Between Tries. n8n's docs note the wait is capped at 5000 ms and tries at five. For longer back-off, use Loop Over Items with a Wait node, or an error workflow that re-queues the item.

What is an n8n error workflow?

An error workflow starts with the Error Trigger node and runs whenever another workflow that names it in its settings fails. It receives the failed workflow's name, the execution URL and the error message, so you can send alerts to email or chat. One error workflow can serve many workflows.

How much memory does n8n need?

It depends on the data each execution holds. n8n keeps execution data in memory, so large datasets, binary files and Code nodes drive usage. Self-hosted instances can raise the Node.js heap limit with NODE_OPTIONS and add RAM; n8n Cloud memory depends on the plan. Batching is the most reliable fix.

Do n8n Cloud webhooks have a timeout?

Yes. n8n's webhook documentation says that on n8n Cloud, if your webhook does not respond within 100 seconds, the incoming request fails with a 524 status, because Cloud runs behind Cloudflare. For long jobs, respond immediately and let the caller poll for the result.

Error handling you can see in the workflow

Autonoly workflows have per-node retries, timeouts, continue-on-fail and error-handler branches, so a failing step can be retried or routed instead of stopping the whole run. Saved workflows have no loop node; lists are worked through by the AI agent or one run per trigger item.