def kill_zombies(self, zombies, session=None):
"""
Fail given zombie tasks, which are tasks that haven't
had a heartbeat for too long, in the current DagBag.
:param zombies: zombie task instances to kill.
:type zombies: airflow.utils.dag_processing.SimpleTaskInstance
:param session: DB session.
:type session: sqlalchemy.orm.session.Session
"""
from airflow.models.taskinstance import TaskInstance # Avoid circular import
for zombie in zombies:
if zombie.dag_id in self.dags:
dag = self.dags[zombie.dag_id]
if zombie.task_id in dag.task_ids:
task = dag.get_task(zombie.task_id)
ti = TaskInstance(task, zombie.execution_date)
# Get properties needed for failure handling from SimpleTaskInstance.
ti.start_date = zombie.start_date
ti.end_date = zombie.end_date
ti.try_number = zombie.try_number
ti.state = zombie.state
ti.test_mode = self.UNIT_TEST_MODE
ti.handle_failure("{} detected as zombie".format(ti),
ti.test_mode, ti.get_template_context())
self.log.info('Marked zombie job %s as %s', ti, ti.state)
session.commit()
def _set_state_to_skipped(self, dag_run, execution_date, tasks, session):
"""
Used internally to set state of task instances to skipped from the same dag run.
"""
task_ids = [d.task_id for d in tasks]
now = timezone.utcnow()
if dag_run:
session.query(TaskInstance).filter(
TaskInstance.dag_id == dag_run.dag_id,
TaskInstance.execution_date == dag_run.execution_date,
TaskInstance.task_id.in_(task_ids),
).update(
{
TaskInstance.state: State.SKIPPED,
TaskInstance.start_date: now,
TaskInstance.end_date: now,
},
synchronize_session=False,
)
else:
if execution_date is None:
raise ValueError("Execution date is None and no dag run")
self.log.warning("No DAG RUN present this should not happen")
# this is defensive against dag runs that are not complete
for task in tasks:
ti = TaskInstance(task, execution_date=execution_date)
ti.state = State.SKIPPED
ti.start_date = now
ti.end_date = now
session.merge(ti)
def skip(self, dag_run, execution_date, tasks, session=None):
"""
Sets tasks instances to skipped from the same dag run.
:param dag_run: the DagRun for which to set the tasks to skipped
:param execution_date: execution_date
:param tasks: tasks to skip (not task_ids)
:param session: db session to use
"""
if not tasks:
return
task_ids = [d.task_id for d in tasks]
now = timezone.utcnow()
if dag_run:
session.query(TaskInstance).filter(
TaskInstance.dag_id == dag_run.dag_id,
TaskInstance.execution_date == dag_run.execution_date,
TaskInstance.task_id.in_(task_ids)).update(
{
TaskInstance.state: State.SKIPPED,
TaskInstance.start_date: now,
TaskInstance.end_date: now
},
synchronize_session=False)
session.commit()
else:
if execution_date is None:
raise ValueError("Execution date is None and no dag run")
self.log.warning("No DAG RUN present this should not happen")
# this is defensive against dag runs that are not complete
for task in tasks:
ti = TaskInstance(task, execution_date=execution_date)
ti.state = State.SKIPPED
ti.start_date = now
ti.end_date = now
session.merge(ti)
session.commit()
def skip(self, dag_run, execution_date, tasks, session=None):
"""
Sets tasks instances to skipped from the same dag run.
:param dag_run: the DagRun for which to set the tasks to skipped
:param execution_date: execution_date
:param tasks: tasks to skip (not task_ids)
:param session: db session to use
"""
if not tasks:
return
task_ids = [d.task_id for d in tasks]
now = timezone.utcnow()
if dag_run:
session.query(TaskInstance).filter(
TaskInstance.dag_id == dag_run.dag_id,
TaskInstance.execution_date == dag_run.execution_date,
TaskInstance.task_id.in_(task_ids)
).update({TaskInstance.state: State.SKIPPED,
TaskInstance.start_date: now,
TaskInstance.end_date: now},
synchronize_session=False)
session.commit()
else:
assert execution_date is not None, "Execution date is None and no dag run"
self.log.warning("No DAG RUN present this should not happen")
# this is defensive against dag runs that are not complete
for task in tasks:
ti = TaskInstance(task, execution_date=execution_date)
ti.state = State.SKIPPED
ti.start_date = now
ti.end_date = now
session.merge(ti)
session.commit()
def _kill_zombies(self, dag, zombies, session):
"""
copy paste from airflow.models.dagbag.DagBag.kill_zombies
"""
from airflow.models.taskinstance import TaskInstance # Avoid circular import
for zombie in zombies:
if zombie.task_id in dag.task_ids:
task = dag.get_task(zombie.task_id)
ti = TaskInstance(task, zombie.execution_date)
# Get properties needed for failure handling from SimpleTaskInstance.
ti.start_date = zombie.start_date
ti.end_date = zombie.end_date
ti.try_number = zombie.try_number
ti.state = zombie.state
# ti.test_mode = self.UNIT_TEST_MODE
ti.handle_failure(
"{} detected as zombie".format(ti),
ti.test_mode,
ti.get_template_context(),
)
self.log.info("Marked zombie job %s as %s", ti, ti.state)
session.commit()
