Hey guys,

I have been successfully using airflow on AWS ECS for a while now. I have
about 30 DAGs that run on a daily basis.
Recently, I noticed that one of my DAGs has been getting stuck in running
state. When I checked the DAG tasks, I noticed that that one of the
downstream tasks just didn't queue up even though it's upstream task was
completed successfully. When I restarted the ECS container, the tasks
continued to queue up and run normally.

I think there is a resource problem where too much memory/cpu is being used
prohibiting the scheduler from performing its duty normally. In summary,
how do I make it so that airflow uses less resources, or that it frees up
resources so the scheduler can run properly?

I am using LocalExecutor by the way on AIrflow 1.8.2.

Thanks!

- Ali

Reply via email to