GitHub user MrHenryD created a discussion: DAGProcessor failing to process dag files after an unknown period of time
Airflow DAGProcessor works fine initially but after a period of time, it fails to parse some DAG files which causes them to be disabled and not scheduled. Having trouble debugging what the issue is and would appreciate any guidance. ## Configuration **Python**: 3.11 **Airflow**: 3.0.2 **Executor**: KubernetesExecutor **Platform**: AWS EKS **Storage**: EFS ## DAG Processing Stats <img width="620" height="118" alt="image" src="https://github.com/user-attachments/assets/eb313a56-1dc8-4d36-af8e-c4b242efaf95"; /> I see errors associated with a particular dag file path, but I don't know how to go about checking for errors (don't know if there's a command I can run to list errors associated with a dag bundle or file). My assumption was that the processor either `lacks resources` or is `timing out`, but based on what I see, this shouldn't be possible. ## Metrics **CPU usage is very low for dag processor (was allocated 2 cpus)** <img width="1090" height="230" alt="image" src="https://github.com/user-attachments/assets/986a1cd2-f44c-44e5-b98c-a0383996217d"; /> **Memory usage is also very low (was allocated 4gb ram)** <img width="1086" height="227" alt="image" src="https://github.com/user-attachments/assets/7e869548-c446-45fa-8bc2-32b0c7d6c512"; /> ## Airflow Settings Timeout settings and refresh interval is already quite high. ``` [dag_processor] bundle_refresh_check_interval = 240 dag_file_processor_timeout = 1800 disable_bundle_versioning = true min_file_process_interval = 600 parsing_processes = 3 print_stats_interval = 300 refresh_interval = 600 stale_bundle_cleanup_interval = 3600 stale_bundle_cleanup_min_versions = 1 stale_dag_threshold = 1800 [scheduler] dag_stale_not_seen_duration = 3600 parsing_cleanup_interval = 600 run_duration = 41460 standalone_dag_processor = True statsd_host = airflow-statsd statsd_on = True statsd_port = 9125 statsd_prefix = airflow ``` GitHub link: https://github.com/apache/airflow/discussions/54274 ---- This is an automatically sent email for [email protected]. To unsubscribe, please send an email to: [email protected]
