rjgoyln opened a new pull request, #71174: URL: https://github.com/apache/airflow/pull/71174
## Summary `KubernetesPodOperator` converted and merged several template-field values in `__init__`, which runs before Jinja rendering. A templated `base_container_name` was copied into the default `container_logs` and never rendered, so log fetching asked for a container that did not exist. A templated `configmaps` string was iterated character by character into `env_from`, and templated `volumes` / `volume_mounts` were rejected at Dag parse time before they could resolve. `__init__` now only stores these arguments; the conversions run in `build_pod_request_obj()` and `container_logs` resolves its base-container fallback lazily. `configmaps` becomes a declared template field, having previously been rendered by accident through the `env_from` list it was merged into. An `env_vars` mapping is the one conversion that still precedes rendering, in a `render_template_fields` override: an env var *name* is a template field of `V1EnvVar`, while a plain dict renders only its values. Only the container shape is inspected, never the value. ## Behavior change `SparkKubernetesOperator` no longer warns that `container_logs` "is not supported and will be overridden" when the user never passed it; an explicitly requested container is still overridden to the driver container. Clears the `KubernetesPodOperator` entry of the burn-down tracked in #70296. --- ##### Was generative AI tooling used to co-author this PR? - [X] Yes — Claude Code (Opus 5) Generated-by: Claude Code (Opus 5) following [the guidelines](https://github.com/apache/airflow/blob/main/contributing-docs/05_pull_requests.rst#gen-ai-assisted-contributions) -- 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]
