I am testing my rss feed from inoreader. I found several issues most probably with n8n cloud infrastructure:
When there are more than 1 simultaneous POST requests to webhook (the same hour+minute+second) - even when raw timstamp differs, there is empty body. I debuged it with webhook.online and there is no issue with body, i also tried differtent action - like google drive file creation from inoreader rss aggregator. All the times it works without issue, confirming that it is something with n8n. More than that, and probably more serious issue - when i do “listen for test event” and more than one rss article arrives to webhook, in the output headers, i can see “leaked” (not mine) inoreader rules (rule to send article to webhook) names and also other users inoreader IDs.
“x-inoreader-rule-name”: “mine inoreader configured name of rule which triggers data sending to n8n webhook”, “other rule name - definitevily not mine”, “other rule name”, “and so on”
“x-inoreader-user-id”: “mine ID”, “other id”, “and so on” - few times
===
So I assume it is either some sort of race condition or load balancer issue
Check N8N Workflow Settings: In N8N, go to your workflow settings (Workflow Settings → Execution Mode). If it is in “Main”, perhaps changing it to “Queue” could help serialize the executions of this specific workflow, but it would not solve the problem if it occurs before the execution starts (in the webhook reception layer). It is worth testing if the Problem (empty body) improves with the “Queue” mode.
Actually i cannot find such settings in the cloud instance (1.90) however it fails even when single node (webhook) is being tested. so i believe changing workflow settings will not change much when first node is webhook and it fails. however i am unable to confirm at 100% as i cannot see that settings in my setup