har*_*han 5 python task command-line-interface directed-acyclic-graphs airflow
I am using Airflow 1.10.2 forked from the popular puckel airflow docker image and using the CLI to operate it reading the documentation. I had created a DAG which is getting executed successfully when I trigger it through the UI. But when I use the command line to trigger the DAG, the DAG status becomes RUNNING but the tasks are never executed.
For references of the images below, the ID of the DAG is xcom_4_modules_ver_5
After running the DAG through a scheduled process and triggering manually through the UI, I executed airflow list_dag_runs xcom_4_modules_ver_5 on CLI to check the status of DAG run. The DAG was successfully executed and wrote the expected output in log file. This corresponds to ID 1 and 2.
However, after executing the command airflow trigger_dag -sd /usr/local/airflow/airflow/dags/ -r CLI-RUN-TEST xcom_4_modules_ver_5 to trigger the DAG from the UI, the status is forever RUNNING. Corresponds to ID 3 in the image below.
I went to the tree view in UI to realise that no component has been executed for very long.
The first two green columns corresponds to run ID 1 and 2 whereas the third column maps to run ID 3.
The airflow scheduler and airflow webserver are on.
Would request Airflow masters here to suggest me on this.
小智 0
这可能不是人们寻求的答案,也不是最适用的答案,但无论如何我都会发布它。Airflow 版本 2 为我们解决了许多 DAG 运行时问题和调度程序堵塞。此前,在 1.10 版本中,这些错误经常发生,但相比之下,Airflow 2 的错误要少得多且更小。但仍然不会说它是一个完美的软件。
如果您决定升级,Apache 有很好的说明可供遵循:https://airflow.apache.org/docs/apache-airflow/stable/upgrading-from-1-10/index.html
如果不是功能更新,以我的体验来看稳定性的提升在一年左右之后就已经是值得的了。
| 归档时间: |
|
| 查看次数: |
235 次 |
| 最近记录: |