Hi, First off, I've only been using Prefect for about two weeks but it's so much better than similar workflows I'd been orchestrating myself in the past. So, thanks to the devs and the active community whose input has helped me tremendously while getting up-to-speed. I'm wondering if anyone has a workaround for this zombie task/concurrency limit issue? It'd be great to see either the underlying bug addressed or a feature added to the UI which allows users to manually cancel/fail/whathaveyou the stuck tasks, but I'll take anything right now as this is looking like it may be a major blocker for me. From within psql, I tried manually setting concurrency_limit.active_slots to '{}' (there were what I assumed to be 20 zombie UUIDs here) but that resulted in the server throwing a bunch of errors when trying to view the "Task Run Concurrency Limits" page in the UI, so I'm guessing there's some callback machinery that doing that circumvents. Any nudges would be greatly appreciated!

Hi Pete. Thanks for raising this one. The issue you raised is not a simple fix (I promise we’re not just ignoring it). I’d recommend using the new global concurrency feature to avoid this issue. If that doesn’t work for you, please tell us as it would be great to understand why/what changes we might need there. Side note- you should be able to reset a task concurrency limit from the CLI or UI.

Thanks, @Jenny. I will have to take a look at the global concurrency feature. That sounds promising, though!