Hello, I was wondering if there's a way to enable debug/trace logs for metaflow? We're running into a scenario where the UI shows all tasks have finished but the step has not. It seems like metaflow is doing something either while executing

self.next

or immediately after that, but before beginning the next step. It would be great to get insights on what might be happening since it runs for a significantly long time in certain cases

interesting. we recently ran into this issue with another deployment where a database connection in the user code hadn't properly terminated leading to this issue. are you seeing this issue across all flows or just a few specific ones?

It happens inconsistently so it's hard to reproduce, specifically in the most recent example it happened with two long running tasks that were fetching data from S3, docker stats during that anomalous period showed high cpu usage (varied from 200 - 7000%), low memory usage (<5%) and no network I/O