Why does the ExternalTaskSensor get stuck?
If you configured an ExternalTaskSensor in Airflow to wait until a task in another DAG finishes (or the entire DAG finishes), you have probably noticed that it behaves surprisingly. Most data engineers assume that it is a bug, and their sensor gets stuck for whatever reason. I have to admit that I find it annoying because the documentation of the ExternalTaskSensor does not warn us about it.
ExternalTaskSensor works only if the other DAG has the same execution date equal to the expected execution date. Sure, you can specify the execution time delta, but it is not a time range.
For example:
- The execution date of the DAG that contains the sensor is: 2020-10-01 14:00:00
- The execution delta = 3 hours
In this situation, the other DAG’s execution date must be 2020-10-01 11:00:00.
I think that most data engineers expect that delta = 3 hours means that the other DAG may execute at any time between 11:00 and 14:00.
Parsing machine learning logs with Ahana, a managed Presto service, and Cube, a headless BI solution

Check out my article published on the Cube.dev blog!
You may also like
- How to use Airflow backfill to run DAGs for a specified date in the past?
- How to check the next execution date of an Airflow DAG
- How to run Airflow in Docker (with a persistent database)
- How to postpone Airflow DAG until files get uploaded into an S3 bucket
- Conditionally pick an Airflow DAG branch using an SQL query
Bartosz Mikulski
- Data/MLOps engineer by day
- DevRel/copywriter by night
- Python and data engineering trainer
- Conference speaker
- Contributed a chapter to the book "97 Things Every Data Engineer Should Know"
- Twitter: @mikulskibartosz