test(e2e): протокол сквозного тестирования и инструменты очистки DWH

- Зачем:
  - формализация проверки всей цепочки ETL (STG -> ODS -> DDS -> DM).
- Что:
  - создан docs/e2e-etl-test-protocol.md и sql/truncate_gp.sql.
  - в Makefile добавлена команда dwh-truncate.
  - start_date во всех DAG изменен на 2017-01-01.
- Проверка:
  - выполнение make dwh-truncate и прогон DAG.

airflow/dags/bookings_dds_ddl.py

@@ -21,7 +21,7 @@ default_args = {"owner": "airflow", "retries": 1, "retry_delay": timedelta(secon
 
 with DAG(
     dag_id="bookings_dds_ddl",
-    start_date=pendulum.datetime(2024, 1, 1, tz="UTC"),
+    start_date=pendulum.datetime(2017, 1, 1, tz="UTC"),
     schedule=None,
     catchup=False,
     template_searchpath="/sql",

airflow/dags/bookings_dm_ddl.py

@@ -24,7 +24,7 @@ default_args = {"owner": "airflow", "retries": 1, "retry_delay": timedelta(secon
 
 with DAG(
     dag_id="bookings_dm_ddl",
-    start_date=pendulum.datetime(2024, 1, 1, tz="UTC"),
+    start_date=pendulum.datetime(2017, 1, 1, tz="UTC"),
     schedule=None,
     catchup=False,
     template_searchpath="/sql",

airflow/dags/bookings_ods_ddl.py

@@ -21,7 +21,7 @@ default_args = {"owner": "airflow", "retries": 1, "retry_delay": timedelta(secon
 
 with DAG(
     dag_id="bookings_ods_ddl",
-    start_date=pendulum.datetime(2024, 1, 1, tz="UTC"),
+    start_date=pendulum.datetime(2017, 1, 1, tz="UTC"),
     schedule=None,
     catchup=False,
     template_searchpath="/sql",

airflow/dags/bookings_stg_ddl.py

@@ -21,7 +21,7 @@ default_args = {"owner": "airflow", "retries": 1, "retry_delay": timedelta(secon
 
 with DAG(
     dag_id="bookings_stg_ddl",
-    start_date=pendulum.datetime(2024, 1, 1, tz="UTC"),
+    start_date=pendulum.datetime(2017, 1, 1, tz="UTC"),
     schedule=None,
     catchup=False,
     template_searchpath="/sql",

airflow/dags/bookings_to_gp_dds.py

@@ -37,7 +37,7 @@ def _finish_summary() -> None:
 
 with DAG(
     dag_id="bookings_to_gp_dds",
-    start_date=pendulum.datetime(2024, 1, 1, tz="UTC"),
+    start_date=pendulum.datetime(2017, 1, 1, tz="UTC"),
     schedule=None,
     catchup=False,
     max_active_runs=1,

airflow/dags/bookings_to_gp_dm.py

@@ -40,7 +40,7 @@ def _finish_summary() -> None:
 
 with DAG(
     dag_id="bookings_to_gp_dm",
-    start_date=pendulum.datetime(2024, 1, 1, tz="UTC"),
+    start_date=pendulum.datetime(2017, 1, 1, tz="UTC"),
     schedule=None,
     catchup=False,
     max_active_runs=1,

airflow/dags/bookings_to_gp_ods.py

@@ -125,7 +125,7 @@ def _finish_summary() -> None:
 
 with DAG(
     dag_id="bookings_to_gp_ods",
-    start_date=pendulum.datetime(2024, 1, 1, tz="UTC"),
+    start_date=pendulum.datetime(2017, 1, 1, tz="UTC"),
     schedule=None,
     catchup=False,
     max_active_runs=1,

airflow/dags/bookings_to_gp_stage.py

@@ -68,7 +68,7 @@ def _finish_summary() -> None:
 
 with DAG(
     dag_id="bookings_to_gp_stage",
-    start_date=pendulum.datetime(2024, 1, 1, tz="UTC"),
+    start_date=pendulum.datetime(2017, 1, 1, tz="UTC"),
     schedule=None,
     catchup=False,
     max_active_runs=1,

airflow/dags/csv_to_greenplum.py

@@ -130,7 +130,7 @@ default_args = {"owner": "airflow", "retries": 1, "retry_delay": timedelta(secon
 
 with DAG(
     dag_id="csv_to_greenplum",
-    start_date=datetime(2024, 1, 1),
+    start_date=datetime(2017, 1, 1),
     schedule=None,
     catchup=False,
     default_args=default_args,

airflow/dags/csv_to_greenplum_dq.py

@@ -52,7 +52,7 @@ default_args = {"owner": "airflow", "retries": 1, "retry_delay": timedelta(secon
 
 with DAG(
     dag_id="csv_to_greenplum_dq",
-    start_date=datetime(2024, 1, 1),
+    start_date=datetime(2017, 1, 1),
     schedule=None,
     catchup=False,
     default_args=default_args,

airflow/dags/ddl_greenplum_base.py

@@ -17,7 +17,7 @@ default_args = {"owner": "airflow", "retries": 1, "retry_delay": timedelta(secon
 
 with DAG(
     dag_id="orders_base_ddl",
-    start_date=datetime(2024, 1, 1),
+    start_date=datetime(2017, 1, 1),
     schedule=None,
     catchup=False,
     template_searchpath="/sql",