Fury0508 commented on issue #70027: URL: https://github.com/apache/airflow/issues/70027#issuecomment-5007070316
Hi, I would be interested in working on this once the expected semantics are agreed. From an initial review, I think there are two possible interfaces: 1. Accept an explicit list of task IDs in the backfill API. 2. Provide a CLI/UI regex or pattern that resolves to that list of task IDs. Before implementing it, I would like to clarify how dependencies and DAG-run state should behave. My proposed initial semantics would be: * Only the selected tasks are scheduled. * Upstream and downstream tasks are not automatically included. * Dependencies on unselected upstream tasks are ignored for this backfill. * The request is rejected when the selection matches no tasks. * The backfill/DAG run is considered successful based only on the selected tasks. * The resolved task IDs are persisted with the backfill so that its behaviour remains reproducible if the DAG later changes. Alternatively, we could automatically include upstream dependencies, but that may not address the sensor use case described in the issue. Would this general direction be acceptable? If so, I would be happy to prepare a PR, beginning with the API/model and tests before adding the CLI and UI support. -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected]
