Greetings! We had a situation yesterday where a runaway API process created several hundred thousand flow runs in a Prefect 2 queue. We couldn’t find a way to delete these in bulk from the Cloud UI, and I could not delete the deployment while the flow runs were active. I ran a process overnight to remove all of the erroneous queued up flow runs, but we’d like to limit the possibility of this happening again. Is there some way to limit the amount of scheduled flow runs from a single deployment that can be active at any time? The queue was a shared one with other deployments so the backlog slowed down scheduled work. I understand this could be avoided by having a separate queue that isolates this type of API connected deployment. Is having queues based on deployment or process ’archetype” (e.g. dbt) best practice? Finally, is there a way in the UI to bulk delete flow runs in a queue based on the deployment name/id?

Hi Scott, there is an open issue for bulk deleting flow runs from the UI. I recommend creating an issue for the configuration of limiting the scheduled flow runs at the deployment level.