Arunodoy18 commented on issue #60271: URL: https://github.com/apache/airflow/issues/60271#issuecomment-3751855992
Thanks for the detailed reproduction and logs. In Airflow 3, worker pods are expected to interact with metadata exclusively via the API server, so direct DB connectivity from workers should not be required. From the stack trace, it appears that the worker container is executing airflow db check, which performs a direct SQLAlchemy connection to the metadata database. This suggests that a DB health check path is still being triggered in the worker context. Looking at the failure timing, this seems to occur during worker startup rather than during task execution, which raises the question of whether this originates from: a worker entrypoint startup check, a shared CLI-based health check reused across components, or a Helm chart–configured lifecycle or probe invoking airflow db check. Before proposing a fix, I’d like to confirm whether this DB check is expected to run in worker pods in Airflow 3, or whether this is an unintended carryover from pre–Airflow 3 behavior. Clarifying this would help determine whether the fix belongs in Airflow core, the KubernetesExecutor implementation, or Helm chart defaults. -- 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]
