def test_backfill_multi_dates(self): dag = self.dagbag.get_dag('example_bash_operator') end_date = DEFAULT_DATE + datetime.timedelta(days=1) executor = MockExecutor() job = BackfillJob( dag=dag, start_date=DEFAULT_DATE, end_date=end_date, executor=executor, ignore_first_depends_on_past=True ) job.run() expected_execution_order = [ ("runme_0", DEFAULT_DATE), ("runme_1", DEFAULT_DATE), ("runme_2", DEFAULT_DATE), ("runme_0", end_date), ("runme_1", end_date), ("runme_2", end_date), ("also_run_this", DEFAULT_DATE), ("also_run_this", end_date), ("run_after_loop", DEFAULT_DATE), ("run_after_loop", end_date), ("run_this_last", DEFAULT_DATE), ("run_this_last", end_date), ] self.assertListEqual( [((dag.dag_id, task_id, when, 1), State.SUCCESS) for (task_id, when) in expected_execution_order], executor.sorted_tasks ) session = settings.Session() drs = session.query(DagRun).filter( DagRun.dag_id == dag.dag_id ).order_by(DagRun.execution_date).all() self.assertTrue(drs[0].execution_date == DEFAULT_DATE) self.assertTrue(drs[0].state == State.SUCCESS) self.assertTrue(drs[1].execution_date == DEFAULT_DATE + datetime.timedelta(days=1)) self.assertTrue(drs[1].state == State.SUCCESS) dag.clear() session.close()
def test_backfill_respect_dag_concurrency_limit(self, mock_log): dag = self._get_dummy_dag('test_backfill_respect_concurrency_limit') dag.concurrency = 2 executor = MockExecutor() job = BackfillJob( dag=dag, executor=executor, start_date=DEFAULT_DATE, end_date=DEFAULT_DATE + datetime.timedelta(days=7), ) job.run() self.assertGreater(len(executor.history), 0) concurrency_limit_reached_at_least_once = False num_running_task_instances = 0 for running_task_instances in executor.history: self.assertLessEqual(len(running_task_instances), dag.concurrency) num_running_task_instances += len(running_task_instances) if len(running_task_instances) == dag.concurrency: concurrency_limit_reached_at_least_once = True self.assertEqual(8, num_running_task_instances) self.assertTrue(concurrency_limit_reached_at_least_once) times_dag_concurrency_limit_reached_in_debug = self._times_called_with( mock_log.debug, DagConcurrencyLimitReached, ) times_pool_limit_reached_in_debug = self._times_called_with( mock_log.debug, NoAvailablePoolSlot, ) times_task_concurrency_limit_reached_in_debug = self._times_called_with( mock_log.debug, TaskConcurrencyLimitReached, ) self.assertEqual(0, times_pool_limit_reached_in_debug) self.assertEqual(0, times_task_concurrency_limit_reached_in_debug) self.assertGreater(times_dag_concurrency_limit_reached_in_debug, 0)
def test_backfill_conf(self): dag = self._get_dummy_dag('test_backfill_conf') executor = MockExecutor() conf_ = json.loads("""{"key": "value"}""") job = BackfillJob(dag=dag, executor=executor, start_date=DEFAULT_DATE, end_date=DEFAULT_DATE + datetime.timedelta(days=2), conf=conf_) job.run() dr = DagRun.find(dag_id='test_backfill_conf') self.assertEqual(conf_, dr[0].conf)
def test_sub_set_subdag(self): dag = DAG('test_sub_set_subdag', start_date=DEFAULT_DATE, default_args={'owner': 'owner1'}) with dag: op1 = DummyOperator(task_id='leave1') op2 = DummyOperator(task_id='leave2') op3 = DummyOperator(task_id='upstream_level_1') op4 = DummyOperator(task_id='upstream_level_2') op5 = DummyOperator(task_id='upstream_level_3') # order randomly op2.set_downstream(op3) op1.set_downstream(op3) op4.set_downstream(op5) op3.set_downstream(op4) dag.clear() dr = dag.create_dagrun(run_id="test", state=State.RUNNING, execution_date=DEFAULT_DATE, start_date=DEFAULT_DATE) executor = MockExecutor() sub_dag = dag.sub_dag(task_regex="leave*", include_downstream=False, include_upstream=False) job = BackfillJob(dag=sub_dag, start_date=DEFAULT_DATE, end_date=DEFAULT_DATE, executor=executor) job.run() self.assertRaises(sqlalchemy.orm.exc.NoResultFound, dr.refresh_from_db) # the run_id should have changed, so a refresh won't work drs = DagRun.find(dag_id=dag.dag_id, execution_date=DEFAULT_DATE) dr = drs[0] self.assertEqual( DagRun.generate_run_id(DagRunType.BACKFILL_JOB, DEFAULT_DATE), dr.run_id) for ti in dr.get_task_instances(): if ti.task_id == 'leave1' or ti.task_id == 'leave2': self.assertEqual(State.SUCCESS, ti.state) else: self.assertEqual(State.NONE, ti.state)
def test_backfill_max_limit_check_within_limit(self): dag = self._get_dag_test_max_active_limits( 'test_backfill_max_limit_check_within_limit', max_active_runs=16) start_date = DEFAULT_DATE - datetime.timedelta(hours=1) end_date = DEFAULT_DATE executor = MockExecutor() job = BackfillJob(dag=dag, start_date=start_date, end_date=end_date, executor=executor, donot_pickle=True) job.run() dagruns = DagRun.find(dag_id=dag.dag_id) self.assertEqual(2, len(dagruns)) self.assertTrue(all(run.state == State.SUCCESS for run in dagruns))
def test_backfill_execute_subdag(self): dag = self.dagbag.get_dag('example_subdag_operator') subdag_op_task = dag.get_task('section-1') subdag = subdag_op_task.subdag subdag.schedule_interval = '@daily' start_date = timezone.utcnow() executor = MockExecutor() job = BackfillJob(dag=subdag, start_date=start_date, end_date=start_date, executor=executor, donot_pickle=True) job.run() subdag_op_task.pre_execute(context={'execution_date': start_date}) subdag_op_task.execute(context={'execution_date': start_date}) subdag_op_task.post_execute(context={'execution_date': start_date}) history = executor.history subdag_history = history[0] # check that all 5 task instances of the subdag 'section-1' were executed self.assertEqual(5, len(subdag_history)) for sdh in subdag_history: ti = sdh[3] self.assertIn('section-1-task-', ti.task_id) with create_session() as session: successful_subdag_runs = ( session .query(DagRun) .filter(DagRun.dag_id == subdag.dag_id) .filter(DagRun.execution_date == start_date) # pylint: disable=comparison-with-callable .filter(DagRun.state == State.SUCCESS) .count() ) self.assertEqual(1, successful_subdag_runs) subdag.clear() dag.clear()
def test_backfill_execute_subdag_with_removed_task(self): """ Ensure that subdag operators execute properly in the case where an associated task of the subdag has been removed from the dag definition, but has instances in the database from previous runs. """ dag = self.dagbag.get_dag('example_subdag_operator') subdag = dag.get_task('section-1').subdag executor = MockExecutor() job = BackfillJob(dag=subdag, start_date=DEFAULT_DATE, end_date=DEFAULT_DATE, executor=executor, donot_pickle=True) removed_task_ti = TI( task=DummyOperator(task_id='removed_task'), execution_date=DEFAULT_DATE, state=State.REMOVED) removed_task_ti.dag_id = subdag.dag_id session = settings.Session() session.merge(removed_task_ti) session.commit() with timeout(seconds=30): job.run() for task in subdag.tasks: instance = session.query(TI).filter( TI.dag_id == subdag.dag_id, TI.task_id == task.task_id, TI.execution_date == DEFAULT_DATE).first() self.assertIsNotNone(instance) self.assertEqual(instance.state, State.SUCCESS) removed_task_ti.refresh_from_db() self.assertEqual(removed_task_ti.state, State.REMOVED) subdag.clear() dag.clear()
def test_backfill_pool_not_found(self): dag = self._get_dummy_dag( dag_id='test_backfill_pool_not_found', pool='king_pool', ) executor = MockExecutor() job = BackfillJob( dag=dag, executor=executor, start_date=DEFAULT_DATE, end_date=DEFAULT_DATE + datetime.timedelta(days=7), ) try: job.run() except AirflowException: return self.fail()
def test_backfill_max_limit_check_complete_loop(self): dag = self._get_dag_test_max_active_limits( 'test_backfill_max_limit_check_complete_loop') start_date = DEFAULT_DATE - datetime.timedelta(hours=1) end_date = DEFAULT_DATE # Given the max limit to be 1 in active dag runs, we need to run the # backfill job 3 times success_expected = 2 executor = MockExecutor() job = BackfillJob(dag=dag, start_date=start_date, end_date=end_date, executor=executor, donot_pickle=True) job.run() success_dagruns = len(DagRun.find(dag_id=dag.dag_id, state=State.SUCCESS)) running_dagruns = len(DagRun.find(dag_id=dag.dag_id, state=State.RUNNING)) self.assertEqual(success_expected, success_dagruns) self.assertEqual(0, running_dagruns) # no dag_runs in running state are left
def test_backfill_run_backwards(self): dag = self.dagbag.get_dag("test_start_date_scheduling") dag.clear() job = BackfillJob(dag=dag, start_date=DEFAULT_DATE, end_date=DEFAULT_DATE + datetime.timedelta(days=1), run_backwards=True) job.run() session = settings.Session() tis = session.query(TI).filter( TI.dag_id == 'test_start_date_scheduling' and TI.task_id == 'dummy').order_by(TI.execution_date).all() queued_times = [ti.queued_dttm for ti in tis] self.assertTrue(queued_times == sorted(queued_times, reverse=True)) self.assertTrue(all([ti.state == State.SUCCESS for ti in tis])) dag.clear() session.close()
def test_trigger_controller_dag(self): dag = self.dagbag.get_dag('example_trigger_controller_dag') target_dag = self.dagbag.get_dag('example_trigger_target_dag') target_dag.sync_to_db() dag_file_processor = DagFileProcessor(dag_ids=[], log=Mock()) task_instances_list = Mock() dag_file_processor._process_task_instances( target_dag, task_instances_list=task_instances_list) self.assertFalse(task_instances_list.append.called) job = BackfillJob(dag=dag, start_date=DEFAULT_DATE, end_date=DEFAULT_DATE, ignore_first_depends_on_past=True) job.run() dag_file_processor._process_task_instances( target_dag, task_instances_list=task_instances_list) self.assertTrue(task_instances_list.append.called)
def test_backfill_rerun_upstream_failed_tasks(self): dag = DAG(dag_id='test_backfill_rerun_upstream_failed', start_date=DEFAULT_DATE, schedule_interval='@daily') with dag: op1 = DummyOperator( task_id='test_backfill_rerun_upstream_failed_task-1', dag=dag) op2 = DummyOperator( task_id='test_backfill_rerun_upstream_failed_task-2', dag=dag) op1.set_upstream(op2) dag.clear() executor = MockExecutor() job = BackfillJob( dag=dag, executor=executor, start_date=DEFAULT_DATE, end_date=DEFAULT_DATE + datetime.timedelta(days=2), ) job.run() ti = TI( task=dag.get_task('test_backfill_rerun_upstream_failed_task-1'), execution_date=DEFAULT_DATE) ti.refresh_from_db() ti.set_state(State.UPSTREAM_FAILED) job = BackfillJob(dag=dag, executor=executor, start_date=DEFAULT_DATE, end_date=DEFAULT_DATE + datetime.timedelta(days=2), rerun_failed_tasks=True) job.run() ti = TI( task=dag.get_task('test_backfill_rerun_upstream_failed_task-1'), execution_date=DEFAULT_DATE) ti.refresh_from_db() self.assertEqual(ti.state, State.SUCCESS)
def test_backfill_examples(self, dag_id, expected_execution_order): """ Test backfilling example dags Try to backfill some of the example dags. Be careful, not all dags are suitable for doing this. For example, a dag that sleeps forever, or does not have a schedule won't work here since you simply can't backfill them. """ dag = self.dagbag.get_dag(dag_id) logger.info('*** Running example DAG: %s', dag.dag_id) executor = MockExecutor() job = BackfillJob(dag=dag, start_date=DEFAULT_DATE, end_date=DEFAULT_DATE, executor=executor, ignore_first_depends_on_past=True) job.run() self.assertListEqual( [((dag_id, task_id, DEFAULT_DATE, 1), State.SUCCESS) for task_id in expected_execution_order], executor.sorted_tasks)
def test_backfill_depends_on_past_backwards(self): """ Test that CLI respects -B argument and raises on interaction with depends_on_past """ dag_id = 'test_depends_on_past' start_date = DEFAULT_DATE + datetime.timedelta(days=1) end_date = start_date + datetime.timedelta(days=1) kwargs = dict( start_date=start_date, end_date=end_date, ) dag = self.dagbag.get_dag(dag_id) dag.clear() executor = MockExecutor() job = BackfillJob(dag=dag, executor=executor, ignore_first_depends_on_past=True, **kwargs) job.run() ti = TI(dag.get_task('test_dop_task'), end_date) ti.refresh_from_db() # runs fine forwards self.assertEqual(ti.state, State.SUCCESS) # raises backwards expected_msg = 'You cannot backfill backwards because one or more tasks depend_on_past: {}'.format( 'test_dop_task') with self.assertRaisesRegex(AirflowException, expected_msg): executor = MockExecutor() job = BackfillJob(dag=dag, executor=executor, run_backwards=True, **kwargs) job.run()
def test_backfill_rerun_failed_tasks_without_flag(self): dag = DAG( dag_id='test_backfill_rerun_failed', start_date=DEFAULT_DATE, schedule_interval='@daily') with dag: DummyOperator( task_id='test_backfill_rerun_failed_task-1', dag=dag) dag.clear() executor = MockExecutor() job = BackfillJob(dag=dag, executor=executor, start_date=DEFAULT_DATE, end_date=DEFAULT_DATE + datetime.timedelta(days=2), ) job.run() ti = TI(task=dag.get_task('test_backfill_rerun_failed_task-1'), execution_date=DEFAULT_DATE) ti.refresh_from_db() ti.set_state(State.FAILED) job = BackfillJob(dag=dag, executor=executor, start_date=DEFAULT_DATE, end_date=DEFAULT_DATE + datetime.timedelta(days=2), rerun_failed_tasks=False ) with self.assertRaises(AirflowException): job.run()