[GitHub] [airflow] dimberman commented on issue #13579: Re-running KubernetesPodOperator task results in AirflowException using Airflow 2.0

[GitBox <gi...@apache.org>]

dimberman commented on issue #13579:
URL: https://github.com/apache/airflow/issues/13579#issuecomment-767736791


   Hi @mfjackson,
   
   So I think there are two different errors here.
   
   The first error is that the Run button is broken for the KubernetesExecutor (as noted in #13805). We should have a fix for that soon.
   
   The second error is specific to the k8sPodOperator.
   
   Basically the "500 bad handshake" return means that kubernetes is trying to speak to a pod that is not currently running. Usually the sidecar container for XCOM stays running until the worker tells it to stop, regardless of task status.
   
   So my guess is for some reason the failed task pod is still up even though it has completed. You might have "is_delete_operator_pod" set to false in either your kubernetespodoperator or in your airflow.cfg. 
   
   I think that deleting the task in the UI, but not deleting the pod in k8s, might be causing this confusion. Can you try deleting the pod and then trying again?


----------------------------------------------------------------
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.

For queries about this service, please contact Infrastructure at:
users@infra.apache.org