@Marvin is there a way to delete persisted parameters after the flow run is completed?

hmm what do you mean by persisted parameters?

The information from the provided excerpts does not directly address the question of deleting persisted parameters after a Prefect flow run is completed. It's possible that this information is not available in the examined documents, or it may be that deletion of persisted parameters is not a commonly documented feature or practice in Prefect. Based on Prefect's design philosophy, parameters are usually defined before the flow run and can be overridden with each run. If you want to remove parameters or any persisted data after a flow run, this typically involves manually deleting them from the storage location (e.g., a database, file system, etc.) or using an automated cleanup task within the flow itself that triggers after the main tasks have completed. If you're looking for more specific guidance on automating this cleanup within Prefect or more details about managing persisted data, I recommend reaching out to the community on the Prefect Slack, or checking the Prefect GitHub Issues for any related discussions or feature requests. You may also go through the official documentation for any updates on this topic as the design and features of Prefect evolve.

@Nate I'm having problems running some flows in a self-hosted solution because I'm passing a array of approximately a thousand strings of filenames as a parameter to a subflow. this value is persisted in the postgres database in the parameters column of the flow run table. and this table is getting very heavy, it has more than 2GB in a table with 50 flowruns. and this is blocking the entire UI because the selects are taking a lot of time.

understood - we have to limit the size of flow run parameters in cloud for this reason, i think its something like half a MB how big in terms of B / KB are your parameters? generally speaking we recommend passing references to large data that flows need instead of the values (partly for this reason)

the biggest flow receive 300 KBs. However i think the serialized strings with the name of the keys of the dics on the json generated from this are much bigger.