7 ways an n8n workflow fails silently, and how to catch each one

The loud failures are easy. A node errors, the run turns red, and you see it. The dangerous ones leave no red mark at all. The client finds out first. Here are seven, with a free way to catch each one inside n8n.

1. Empty results that pass as success

What happens. A search, a database query or an API call returns zero items. Every later node has nothing to do, so the run ends green. The client's leads, orders or invoices never got processed.

Catch it for free. Turn on Always Output Data for the node that can come back empty. Then add an IF node that checks whether the result is empty, and send yourself a message on that branch. Our free Empty run guard template sets this up.

With RunLookout. You pick a step in the workflow. If a run finishes "successfully" but that step produced 0 items, you get a Telegram alert.

2. A schedule that stops firing

What happens. A workflow with a Schedule Trigger quietly stops running. Nothing errors, because nothing runs. You only notice when the daily report is missing.

Catch it for free. Add a heartbeat. At the end of the workflow, ping a free dead man's switch service. If the ping stops, the service tells you. You can also check the execution list for the expected run times.

With RunLookout. We compare each scheduled workflow with its runs. If one is missed by more than a 10-minute grace, you get an alert.

3. A workflow that got deactivated

What happens. Someone turns a workflow off while testing and forgets to turn it back on. Triggers stop. No error appears.

Catch it for free. Open the workflow list on a set day each week and check the Active toggle on the ones that matter. Keep a short list of workflows that must always be on.

With RunLookout. A deactivated scheduled workflow stops producing runs, so it shows up as a missed run.

4. Expired credentials

What happens. A token expires or a password changes. The next run fails at the node that uses it. If nobody is watching executions, the failure sits there for days.

Catch it for free. Create a workflow that starts with the Error Trigger node. In the settings of every other workflow, set it as the Error Workflow. When any run fails, the Error Trigger workflow can send you an email or a chat message.

With RunLookout. The failed run triggers an alert within about 5 minutes, with a plain-English note you can forward to the client.

5. An error swallowed by Continue On Fail

What happens. You set a node to keep going when it errors. The error is real, but the run still ends green. The Error Trigger never fires, because the run did not fail.

Catch it for free. After the node, add an IF node that checks the output for an error field. Send an alert from the error branch. Use the continue-with-error-output option, where available, so failed items go to their own branch.

With RunLookout. If the swallowed error leaves a chosen step with 0 items, the empty run check flags it.

6. Executions that are not saved, so there is nothing to check

What happens. n8n can skip saving successful or failed runs, through the workflow settings or the server's settings. With no saved runs, you cannot look back and see what happened. Any tool that reads execution history has nothing to read.

Catch it for free. In each workflow's settings, set Save successful production executions and Save failed production executions to save. Check the instance's data pruning settings so history stays long enough.

With RunLookout. We read runs through the n8n API, so saved runs are what we check. Keep saving turned on for the workflows you want watched.

7. A server or webhook that goes down

What happens. The n8n server stops, restarts badly or loses its network. Webhooks return errors to whoever calls them. No workflow can report its own failure, because nothing is running.

Catch it for free. Use an uptime monitor outside the server. Point it at your n8n address, and at one webhook URL if the client depends on it. The monitor must run somewhere other than the n8n server.

With RunLookout. We check each server every 5 minutes. If it cannot be reached two checks in a row, you get an alert.

Where to start

Do these three first: set an Error Workflow, add a heartbeat to scheduled workflows, and guard every step that can return nothing. That covers most silent failures.

Let RunLookout watch for you

RunLookout checks your client servers every 5 minutes through the n8n API. It only reads. It never edits a workflow. Alerts arrive on Telegram. The beta is free for 14 days. After that, Solo is $9a month for up to 10 client servers, and Unlimited is $19 a month.