Interesting. If you haven't already, can you create a Jira and append an
example dag that I can run to reproduce (likely capturing the code you have
above). You can then assign the bug to me to look into.
Also, please provide enough context on your use-case and why you are
structuring your code this way. It will help identify any alternatives.
It's not clear to me what you want to do exactly.
-s
On Tue, Aug 2, 2016 at 8:02 PM, David Klosowski <dav...@thinknear.com>
wrote:
> start_date being updated isn't the issue here. I haven't changed it. New
> execution_dates keep getting created for the past before any dags or
> start_dates existed.
>
> Cheers,
> David
>
> On Tue, Aug 2, 2016 at 7:10 PM, siddharth anand <san...@apache.org> wrote:
>
> > The problem might be that the start_date does not get updated. I work
> > around this by changing the name of my dag. I do lose history as well,
> but
> > it works.
> >
> > My dags are named "some_dag_v1". When I change a start date, I update the
> > version suffix to force a reload : "some_dag_v2"
> >
> > -s
> >
> > On Tue, Aug 2, 2016 at 6:49 PM, David Klosowski <dav...@thinknear.com>
> > wrote:
> >
> > > I have a DAG that I just deployed that the scheduler keeps scheduling
> for
> > > the last two months in the past.
> > >
> > > start_date: 8/5/2016
> > >
> > > scheduled runs started:
> > > 7/3/2016
> > > 6/5/2016
> > >
> > > Here is the gist of this DAG's architecture:
> > >
> > > The DAG depends another dags tasks using 7 dynamic ExternalTaskSensors
> > that
> > > it builds which that represent 'daily' jobs and then has a
> DummyOperator
> > > task which aggregates and triggers the 'weekly' job task upon
> completion.
> > >
> > > Some of the code showcasing this:
> > >
> > > run_for_date = datetime(2016, 8, 2)
> > >
> > > args = {'owner': 'airflow',
> > > 'depends_on_past': False,
> > > 'start_date': run_for_date,
> > > 'email': [alert_email],
> > > 'email_on_failure': True,
> > > 'email_on_retry': False,
> > > 'retries': 1,
> > > 'trigger_rule' : 'all_success'}
> > >
> > > dag = DAG(dag_id='weekly_no_track', default_args=args,
> > > schedule_interval=timedelta(days=7),
> > > max_active_runs=1)
> > >
> > >
> > > downstream_task = dag.get_task('wait-for-dailies')
> > > for weekday in [MO, TU, WE, TH, FR, SA, SU]:
> > > task_id = 'wait-for-daily-{day}'.format(day=weekday)
> > >
> > > # weekday(-1) subtracts 1 relative week from the given weekday,
> > however
> > > if the calculated date is already Monday,
> > > # for example, -1 won't change the day.
> > > delta = relativedelta(weekday=weekday(-1))
> > >
> > > sensor = ExternalTaskSensor(task_id=task_id, dag=dag,
> > > external_dag_id='daily_no_track',
> > > external_task_id='daily-no-track',
> > > execution_delta=delta, timeout=86400)
> #
> > > 86400 = 24 hours
> > > sensor.set_downstream(downstream_task)
> > >
> > >
> > > I don't understand what is going on. Why is the scheduler doing
> this? I
> > > want the DAG to start considering dates from today and on in UTC.
> > >
> > > Cheers,
> > > David
> > >
> >
>
