Logs from the airflow container are logged twice #473
Description
Logs from the airflow container are logged twice.
They are correctly written with the logging framework to the file airflow.py.json:
{
"cluster": "airflow",
"container": "airflow",
"file": "airflow.py.json",
"level": "INFO",
"logger": "flask_appbuilder.baseviews",
"message": "Registering route / ('GET',)",
"namespace": "kuttl-test-heroic-badger",
"pod": "airflow-scheduler-default-0",
"role": "scheduler",
"roleGroup": "default",
"timestamp": "2024-07-15T17:53:18.924Z"
}Additionally, the standard output is incorrectly captured (including the ANSI escape codes) and written to the file container.stdout.log:
{
"cluster": "airflow",
"container": "airflow",
"file": "container.stdout.log",
"level": "INFO",
"logger": "ROOT",
"message": "[\u001b[34m2024-07-15T17:53:18.924+0000\u001b[0m] {\u001b[34mbaseviews.py:\u001b[0m319} INFO\u001b[0m - Registering route / ('GET',)\u001b[0m",
"namespace": "kuttl-test-heroic-badger",
"pod": "airflow-scheduler-default-0",
"role": "scheduler",
"roleGroup": "default",
"timestamp": "2024-07-15T17:53:19.882505453Z"
}This problem was introduced in #311 in airflow_controller.rs#L748-L752.
The shell output must not be captured anymore but it must also be ensured that the logs of the Kubernetes executors are still gathered.