sjmiller609 edited a comment on issue #6067: [AIRFLOW-5448] Handle istio-proxy for Kubernetes Executor URL: https://github.com/apache/airflow/pull/6067#issuecomment-530091280 > I would like to add that there is only one difference between KubernetesPodOperator and GKEPodOperator. GKEPodOperator provides authorization for the cluster and then shares all logic with KubernetesPodOperator. Oh perfect, I did not realize that. Also, I now understand that you are attempting to reproduce the original problem in order to make regression tests. I am interested to know where you would put these tests, so that I might contribute such tests in the future if I do something like this. To check if the problem is occurring in a current version of Airflow, you should look from the kubernetes' perspective. I diagnose it by using airflow on kubernetes with the kubernetes executor. Then I make a DAG using the bash operator, and set the dag to run. I watch the task begin / the pod for the task being created with the command ```kubectl get pods --all-namespaces``` I can see the task come up, and kubernetes will show the pod at 1 more container than usual (in my case, two containers - the task container and istio-proxy). The task will complete, then the container goes down. The issue is that the pod is not cleaned up because the other container, istio-proxy does not stop. That can be witnessed with another ```kubectl get pods --all-namespaces```, where you will see the task pod at 1/2 containers (because the task stopped but istio-proxy was not cleaned up)
---------------------------------------------------------------- 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: us...@infra.apache.org With regards, Apache Git Services
