topic Re: Concurrent Workflow Jobs in Data Engineering

Concurrent Workflow Jobs

holychs — Thu, 05 Dec 2024 16:02:16 GMT

Hi Community,

I am trying to run a Databricks workflow job using run_job_task under a for_loop. I have set the concurrent jobs as 2.

I can see 2 iteration jobs getting triggered successfully. But both fail with an error:

"ConnectException: Connection refused (Connection refused)

Error while obtaining a new communication channel"

Contradictory to the fact that same job gets completed successfully with concurrency as 1.

Cluster Configuration: Driver: Standard_DS5_v2 · Workers: Standard_DS5_v2 · 4-8 workers

Re: Concurrent Workflow Jobs

Takuya-Omi — Thu, 05 Dec 2024 16:53:59 GMT

Hi, @holychs

Did you encounter any error messages related to an OOM (Out of Memory) error?
It’s possible that the driver node of the cluster doesn’t have sufficient resources (CPU, memory) to handle multiple concurrent jobs.

Re: Concurrent Workflow Jobs

holychs — Thu, 19 Dec 2024 07:49:32 GMT

It was an internal bug resolved with managing different parameters for each loop jobs.