Replies: 1 comment
-
solved by override clear_xcom_data |
Beta Was this translation helpful? Give feedback.
0 replies
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
-
Original Problem:
Databricks occasionally returns a 500 error on the get_run_id status request. The plan is to use XCom to retrieve the previous run_id on the second trial and check if the first trial is still running. If it is, the run_id will be returned when the second trial submits the run.
Issue with the Above Solution:
The current task is configured with 3 retries, so when the first attempt fails, the second attempt is triggered.
I applied do_xcom_push on the DatabricksSubmitRunOperator to ensure that the run_id is saved in XCom when it is submitted successfully. I also verified via query that the latest XCom value is saved. However, when I use xcom_pull, session.query(XCom), or connector.exec to retrieve the value, I only get the XCom value from the prior date (the last retry of the previous run), not the XCom value from the first trial of the current run.
Beta Was this translation helpful? Give feedback.
All reactions