Occasionally, Airflow DAGs get stuck in the running state but don’t want to run any tasks. From my observations, it happens mostly when we clear many DAG runs in one DAG because we want to reprocess a large number of tasks.
Table of Contents
To avoid this problem, I suggest using the backfill command to restart only a small subset of DAGS. For example, a week of work, wait until it finishes, and restarting another batch.
Anyway, what to do when it is too late, and we have already ended up in a situation when everything is running, but nothing actually wants to run.
The first thing we can do is using the airflow clear command to remove the current state of those DAG runs. We can specify the date range using the -s and -e parameters:
airflow clear -s 2020-01-01 -e 2020-01-07 dag_id
Get Weekly AI Implementation Insights
Join engineering leaders who receive my analysis of common AI production failures and how to prevent them. No fluff, just actionable techniques.
When that is not enough, we need to use the Airflow UI. In the menu, click the ‘Browse” tab, and open the ‘DAG Runs’ view.
On this page, we should find the DAG runs that don’t want to run, select them, and click the ‘With selected’ menu option. In the new menu, we click the ‘Delete’ command.
After that, Airflow should recreate the missing task instances and perhaps starts to execute the code.