@Marvin: is there a timeout on flow running for a long time?

It seems there is indeed a consideration for timeouts or handling long-running flows in Prefect, and from the issues in the GitHub repository, we can draw some insights: 1. Issue #13060: This indicates there can be a timeout set up for tasks within a flow. If tasks exceed this timeout, the worker monitoring the task might die but the job could still be running on the backend (e.g., GCP). This means there are timeouts that can be set, but their enforcement might need additional handling. Worker dies because timeout is not respected. 2. Issue #12864: This issue discusses handling subflow runs that can outlive their parent flows, requiring manual cancellation if the parent flow fails. This suggests that long-running subflows need explicit handling to ensure they are properly managed. Support subflow cancellation if parent flow is in terminal state. 3. Issue #12688: This issue deals with the handling (or mishandling) of cancellations, particularly with long-running tasks, and how Prefect deals with these states. In certain cases, the proper cleanup or state transition on cancellation might not occur as expected. Unable to handle 'Cancelling' events. From these, it is evident that timeouts and handling long-running flows and tasks are supported but might require proper setup and may sometimes face issues, particularly around cleanup and state handling. You can look into more details and specific configurations for setting up these timeouts in the Prefect documentation under Task and Flow configuration. For more specific examples or guidance, feel free to ask more targeted questions.