Scheduler skipped DAG execution?

[Chris Riccomini]

Hey all,

I am observing some weird scheduling behavior. I have a DAG with this schedule defined:

dag = DAG(
    'my-dag',
    default_args=default_args,
    # Run on the 3rd of every month at 12pm UTC.
    schedule_interval='0 12 3 * *',
    # Disable parallelism. The table and file are the same across executions.
    max_active_runs=1)

I was expecting it to run at 12 UTC (4am PST). It did not. The schedule has been running the whole time. The logs in the scheduler show this periodically (every minute or so):

[2016-02-03 16:44:29,047] {models.py:124} INFO - Filling up the DagBag from /etc/airflow/dags
[2016-02-03 16:44:29,048] {models.py:197} INFO - Importing /etc/airflow/dags/airflow-dags/my_dag/my_dag.py
[2016-02-03 16:44:29,057] {models.py:284} INFO - Loaded DAG <DAG: my_dag>
[2016-02-03 16:44:29,058] {models.py:197} INFO - Importing /etc/airflow/dags/airflow-dags/hello_world/hello_world.py
[2016-02-03 16:44:29,065] {models.py:284} INFO - Loaded DAG <DAG: hello-world>
[2016-02-03 16:44:29,066] {models.py:197} INFO - Importing /etc/airflow/dags/weflow/hooks/gc_base_hook.py
[2016-02-03 16:44:29,088] {models.py:197} INFO - Importing /etc/airflow/dags/airflow-dags/hello_world/hello_world.py
[2016-02-03 16:44:29,095] {models.py:284} INFO - Loaded DAG <DAG: hello-world>

And then on a periodic basis (every 5s, or so), I just see this:

[2016-02-03 16:48:29,045] {jobs.py:507} INFO - Prioritizing 0 queued jobs
[2016-02-03 16:48:29,063] {models.py:197} INFO - Importing /etc/airflow/dags/airflow-dags/hello_world/hello_world.py
[2016-02-03 16:48:29,070] {models.py:284} INFO - Loaded DAG <DAG: hello-world>
[2016-02-03 16:48:29,134] {jobs.py:455} INFO - Getting list of tasks to skip for active runs.
[2016-02-03 16:48:29,134] {jobs.py:470} INFO - Checking dependencies on 0 tasks instances, minus 0 skippable ones
[2016-02-03 16:48:29,244] {jobs.py:455} INFO - Getting list of tasks to skip for active runs.
[2016-02-03 16:48:29,245] {jobs.py:470} INFO - Checking dependencies on 0 tasks instances, minus 0 skippable ones
[2016-02-03 16:48:29,295] {jobs.py:633} INFO - Done queuing tasks, calling the executor's heartbeat
[2016-02-03 16:48:29,295] {jobs.py:636} INFO - Loop took: 0.257886 seconds

I'm just running the local executor. The UI also reflects that the DAG is showing up properly. Why wasn't my-dag run?

Cheers,

Chris

[Chris Riccomini]

These are my default_args:

default_args = {
    'owner': 'foo',
    'depends_on_past': False,
    'start_date': datetime(2016, 1, 12),
    'email': ['f...@bar.com'],
    'email_on_failure': True,
    'email_on_retry': False,
    'retries': 1,
    'retry_delay': timedelta(minutes=10),
    'bigquery_conn_id': 'bigquery',
    'google_cloud_storage_conn_id': 'google_cloud_storage',
}

[Maxime Beauchemin]

answer from gitter: