Posts

Showing posts with the label n8n long running task workaround

n8n Webhook Timeouts: Stop Duplicate Runs and Botnet Spam

Image
 Last month, when I faced a sudden wave of 504 gateway timeouts and dropped webhooks on my automation cluster, I realized the massive gap between clean documentation and brutal production reality.  When third-party APIs or upstream servers drop unexpectedly, default retry settings flood and jam queues. I learned the hard way that running automation without proper exponential backoff and dead-letter queues is like setting your own data pipeline on fire. Based on those scars, and on patterns that show up again and again in n8n's own community forum, this guide breaks down how to protect your infrastructure before the next network disruption hits. The timing matters too. Google just led a takedown of the XCPE botnet, a network built from more than 1.3 million hijacked Android TV boxes that a device manufacturer allegedly infected with malware before the units ever left the factory. Takedowns like that do not make scanning traffic vanish overnight.  If your n8n instance faces...

n8n Webhook Timeout Workaround for Long-Running Tasks

Image
Your n8n workflow calls a local LLM through Ollama. The model needs 95 seconds to generate a response. The caller gets a 504 Gateway Timeout anyway, even though the workflow is still running fine in the background. This is one of the most common production issues in self-hosted n8n. It rarely shows up in simple automations. It shows up the moment you add real processing time: PDF parsing, large dataset transforms, API calls that chain three or four services, or a local AI model call that takes longer than a page load. The webhook did not fail. The workflow did not fail. The proxy sitting in front of n8n gave up waiting, and the caller has no way to know the job is still running. If you are searching for an n8n workflow timeout after 120 seconds fix, the short version is this: stop holding an HTTP connection open for the full length of a long job. Respond fast, process in the background, and report the result somewhere else. The rest of this guide shows exactly how to build that, with c...