johnhoran opened a new issue, #58490:
URL: https://github.com/apache/airflow/issues/58490

   ### Apache Airflow Provider(s)
   
   cncf-kubernetes
   
   ### Versions of Apache Airflow Providers
   
   When using KubernetesPodOperator in deferred mode I recently ran into an 
issue where airflow wasn't able to hand back from the triggerer to the worker.  
The underlying pod had succeeded but the task failed, and subsequent retries 
failed for the same reason until the task was manually cleared.   I suspect 
there is some underlying bug here, that I'm not currently able to reproduce, 
but I think it raises another point.  
   
   KPO doesn't actually try recreate the pod on task failure, it tries to adopt 
the existing pod.  If the existing pod has succeeded by the time we get to the 
retry, then why defer at all, can we just go straight to the reentry method?
   
   ### Apache Airflow version
   
   2.11
   
   ### Operating System
   
   linux
   
   ### Deployment
   
   Official Apache Airflow Helm Chart
   
   ### Deployment details
   
   _No response_
   
   ### What happened
   
   _No response_
   
   ### What you think should happen instead
   
   _No response_
   
   ### How to reproduce
   
   -
   
   ### Anything else
   
   _No response_
   
   ### Are you willing to submit PR?
   
   - [x] Yes I am willing to submit a PR!
   
   ### Code of Conduct
   
   - [x] I agree to follow this project's [Code of 
Conduct](https://github.com/apache/airflow/blob/main/CODE_OF_CONDUCT.md)
   


-- 
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.

To unsubscribe, e-mail: [email protected]

For queries about this service, please contact Infrastructure at:
[email protected]

Reply via email to