Deploying n8n on Kubernetes with PostgreSQL as the database, an “On form submission” workflow was created under a single n8n instance, which successfully initiated the process and displayed the form in a pop-up. However, when scaling n8n to multiple instances, the process consistently stalled, and the endpoint /form-test/52aba57b-a4b7-40bf-a81d-ac5cdcb86a5f returned a 404 error. The logs showed the following error:
Received request for unknown webhook: The requested webhook “52aba57b-a4b7-40bf-a81d-ac5cdcb86a5f” is not registered. {“file”:“webhook-request-handler.js”,“function”:“handleRequest”}
2026-01-16T04:03:29.858Z | debug | Querying database for waiting executions {“scopes”:[“waiting-executions”],“file”:“wait-tracker.js”,“function”:“getWaitingExecutions”}
Please help confirm the reason for this.
This is usually because the workflow is not set to Active (< version 2.0) or not published (v 2+). In the Edit Workflow view, click on Publish to make the endpoint active (or flick the Active switch in previous versions).
The currently deployed n8n version number is 2.2.1. I tried using the latest 32.3.5 but still encountered this issue. Strangely, deploying a single instance can work normally, but there are problems with multiple instances
Received request for unknown webhook: The requested webhook “8e6d6c40-a9e9-4d06-9676-e9650e2b7a16” is not registered. {“currentlyRegistered”:[“GET 8e6d6c40-a9e9-4d06-9676-e9650e2b7a16”,“POST 8e6d6c40-a9e9-4d06-9676-e9650e2b7a16”,“GET 47b45936-8761-4467-b338-9f93be28d7ab”,“POST 47b45936-8761-4467-b338-9f93be28d7ab”],“file”:“webhook-request-handler.js”,“function”:“handleRequest”}
The error message in the log is as shown above. This error message indicates that there are two ways to register a webhook: 8 e6d6c40-a9e9-4d06-9676-e9650e2b7a16、47b45936-8761-4467-b338-9f93be28d7ab, Among them, 8e6d6c40-a9e9-4d06-9676-e9650e2b7a16 is already in the registration list. Please help confirm if there are any other reasons that may cause the database to be inconsistent with the query results
I roughly know the reason why deploying multiple instances on k8s requires combining queue processing for webhooks. Supporting queue mode requires an enterprise version, while the community office can currently only support a single instance