I’ve got some questions about human in the loop:
-
How long does the automation stay in “Waiting” - so how much time does the human have to respond? Is there even a time limit? I am currently testing and it’s definitely over 5 minutes
-
Is there a limit as to how many workflows can be “Waiting for feedback from human” at the same time?
Realististic example: Let’s say I want to send out 50-100 human in the loop to different employees once a week. I want to give them 1-2 days to respond. I except around 50% of them not to respond. That would also mean that after a few months I’d have hundreds of “waiting” workflows (unless they crash after a while themselves) - Is what I have in mind doable somehow?
(I’m on the cloud plan - but I’m also interested in how things change for self hosted)
Max wait time?
n8n Docs – Wait Node mentions that the Wait node keeps execution paused until a condition (such as an external signal) is met.
Concurrent waiting workflows limit?
There is no explicit documented limit, but n8n Cloud Plans specify execution retentions and resource limits (storage and active executions).The real limitation comes from database space and overall performance. In n8n Cloud, this is handled automatically.
100 HITL per week long term
Users have discussed HITL scalability in the community, and similar cases have been implemented with subflow splitting and cleanup strategies. Recommendations include using Wait nodes, Webhooks, and expiration handling with If or timeout control.
Cloud vs Self-hosted
The n8n Docs for Self-hosted detail how you can scale infrastructure, change databases (PostgreSQL), and use queue mode for high volume. n8n Cloud has plan-defined limits (e.g., number of active executions and data retention) visible in the user interface.
So you’re saying if I’m in the cloud, and I implement a timeout handling (lets say after 5 days) then I should be fine?