1fanwang opened a new pull request, #73745: URL: https://github.com/apache/airflow/pull/73745
Operators running `airflow db clean` cannot prune `rendered_task_instance_fields` (RTIF) independently of `task_instance`. Because RTIF tables store full rendered templates, configuration payloads, and Kubernetes pod specs for every task attempt, they grow rapidly into gigabytes of database storage while task instances are retained for audit. `airflow db clean --tables rendered_task_instance_fields` fails with `No tables selected for db cleanup`. ### What changed This adds support for cleaning child tables that lack an independent recency timestamp by allowing `_TableConfig` to join a parent recency table on primary key columns. `rendered_task_instance_fields` is configured to join `task_instance` on composite keys `(dag_id, task_id, run_id, map_index)` and filter by `task_instance.start_date`, enabling independent RTIF pruning while retaining task instances. --- ##### Was generative AI tooling used to co-author this PR? - [X] Yes Generated-by: GitHub Copilot CLI following [the guidelines](https://github.com/apache/airflow/blob/main/contributing-docs/05_pull_requests.rst#gen-ai-assisted-contributions) --- ### Testing Done #### Before (main branch) Attempting to clean `rendered_task_instance_fields` on the main branch fails because the table is unregistered: ```bash airflow db clean --tables rendered_task_instance_fields --clean-before-timestamp 2026-01-01 -y ``` ``` SystemExit: No tables selected for db cleanup. Please choose valid table names. ``` #### After (this branch) Executed database setup with 1 task instance and 1 rendered task instance field row: ```bash python /tmp/reproduce_rtif_cleanup.py ``` ``` Seed complete: inserted 1 task_instance and 1 rendered_task_instance_fields row. ``` CLI cleanup purges expired RTIF rows into the archive table while leaving parent task instances intact: ```bash airflow db clean --tables rendered_task_instance_fields --clean-before-timestamp 2026-01-01 -y ``` ``` Checking table rendered_task_instance_fields Found 1 rows meeting deletion criteria. Performing Delete... Moving data to table _airflow_deleted__rendered_task_instance_fields__20260926014710 Finished Performing Delete ``` Inspected post-clean row counts: ```bash python -c " from sqlalchemy import func, select from airflow.models import TaskInstance from airflow.models.renderedtifields import RenderedTaskInstanceFields from airflow.utils.session import create_session with create_session() as session: ti_cnt = session.scalar(select(func.count(TaskInstance.id)).where(TaskInstance.dag_id == 'rtif_demo')) rtif_cnt = session.scalar(select(func.count(RenderedTaskInstanceFields.dag_id)).where(RenderedTaskInstanceFields.dag_id == 'rtif_demo')) print(f'task_instance remaining: {ti_cnt}') print(f'rendered_task_instance_fields remaining: {rtif_cnt}') " ``` ``` task_instance remaining: 1 rendered_task_instance_fields remaining: 0 ``` <details> <summary>Reproducer source: reproduce_rtif_cleanup.py</summary> ```python #!/usr/bin/env python3 import pendulum from airflow.models import DAG, DagModel, DagRun, TaskInstance from airflow.models.dag_version import DagVersion from airflow.models.dagbundle import DagBundleModel from airflow.models.renderedtifields import RenderedTaskInstanceFields from airflow.models.serialized_dag import SerializedDagModel from airflow.providers.standard.operators.python import PythonOperator from airflow.serialization.serialized_objects import LazyDeserializedDAG from airflow.utils.session import create_session from airflow.utils.types import DagRunType base_date = pendulum.DateTime(2023, 1, 1, tzinfo=pendulum.timezone("UTC")) bundle_name = "dags" with create_session() as session: session.add(DagBundleModel(name=bundle_name)) session.flush() dag_id = "rtif_demo" dag = DAG(dag_id=dag_id) dag.fileloc = __file__ dm = DagModel(dag_id=dag_id, bundle_name=bundle_name) session.add(dm) SerializedDagModel.write_dag(LazyDeserializedDAG.from_dag(dag), bundle_name=bundle_name) dag_version = DagVersion.get_latest_version(dag.dag_id) dr = DagRun(dag.dag_id, run_id="run_1", run_type=DagRunType.SCHEDULED, start_date=base_date) ti = TaskInstance( PythonOperator(task_id="task_1", python_callable=print), run_id=dr.run_id, dag_version_id=dag_version.id, ) ti.dag_id = dag.dag_id ti.start_date = base_date session.add(dr) session.add(ti) session.flush() rtif = RenderedTaskInstanceFields(ti=ti, render_templates=False, rendered_fields={"sql": "SELECT 1"}) session.add(rtif) session.commit() print("Seed complete: inserted 1 task_instance and 1 rendered_task_instance_fields row.") ``` </details> --- * Read the **[Pull Request Guidelines](https://github.com/apache/airflow/blob/main/contributing-docs/05_pull_requests.rst#pull-request-guidelines)** for more information. Note: commit author/co-author name and email in commits become permanently public when merged. * For fundamental code changes, an Airflow Improvement Proposal ([AIP](https://cwiki.apache.org/confluence/display/AIRFLOW/Airflow+Improvement+Proposals)) is needed. * When adding dependency, check compliance with the [ASF 3rd Party License Policy](https://www.apache.org/legal/resolved.html#category-x). * For significant user-facing changes create newsfragment: `{pr_number}.significant.rst`, in [airflow-core/newsfragments](https://github.com/apache/airflow/tree/main/airflow-core/newsfragments). You can add this file in a follow-up commit after the PR is created so you know the PR number. -- 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]
