We're trying to migrate some analytics report generation to CrewAI. Scheduled crews keep failing with process timeout errors after exactly 60 minutes. This is a production blocker.
Our setup:
* Crew runs daily, 5-6 agents, uses LLM (OpenAI GPT-4).
* Task chain is sequential. Some tasks involve data processing before the LLM call.
* We're on the latest version. No custom timeouts set in our crew/kickoff configuration.
What we've tried:
* Isolated the slowest agent task. It completes fine standalone (<10 mins).
* Checked for async/await issues. Everything seems standard.
* No clear errors in logs besides the timeout.
Questions:
* Is there a hidden default global timeout in CrewAI's scheduler?
* Are there known issues with process isolation for long-running crews?
* What specific config or decorator actually controls this, since the docs are vague?
Vanity metrics like "agents created" are useless. I need the actual levers to control process execution time.
If it's not a retention curve, I don't care.