loozhengyuan opened a new issue #10048: URL: https://github.com/apache/airflow/issues/10048
In the [docstring](https://github.com/apache/airflow/blob/59cbff0874dd5318cda4b9ce7b7eeb1aad1dad4d/airflow/providers/cncf/kubernetes/operators/kubernetes_pod.py#L47) of the `KubernetesPodOperator`, it is noted that: > If you use Google Kubernetes Engine, use `GKEPodOperator`, which simplifies the authorization process. I think this is rather misleading if you deployed Airflow on GKE. If Airflow is deployed on GKE, one can simply use the default `KubernetesPodOperator` and pass `in_cluster=True`. In contrast, to use the `GKEPodOperator`, I will need to store/pass variables required by `gcloud container clusters get-credentials` to `GKEPodOperator`, i.e. variables like `project_id`, `location`, `cluster_name`, etc. Also, if one is using the default [Airflow Docker image](https://hub.docker.com/r/apache/airflow), it also does not come with `gcloud` and `kubectl` installed, so one may need to build their own Docker image in order to call `GKEPodOperator`. As such, I think the note should be further refined to read: > If you use Google Kubernetes Engine but deployed Airflow outside of the cluster, use `GKEPodOperator`, which simplifies the authorization process. Of course, we could also add another note to mention about the `in_cluster=True` option, but I think I will stick to my recommendation above for now. Comments are welcome 😄 ---------------------------------------------------------------- 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: [email protected]
