dolphinscheduler1.3.1 submit spark task with spark-cluster model，yarn log success，but dolphinscheduler task log show failure。

[lishiyucn]

基础环境： spark 2.3.4 dolphinscheduler1.3.1 hdp3.1.0

使用dolphinscheduler1.3.1调度spark任务使用cluster模式，任务是正常执行成功的，但是dolphinscheduler现实任务执行失败。

dolphinscheduler的状态 一个图标× -解释为 失败

spark日志的状态： [INFO] 2020-09-24 03:50:41.140 - [taskAppId=TASK-3-12-12]:[121] - -> 20/09/24 03:50:41 INFO Client: client token: N/A diagnostics: N/A ApplicationMaster host: 10.2.12.3 ApplicationMaster RPC port: 0 queue: default start time: 1600919407936 final status: SUCCEEDED tracking URL: http://test-2:8088/proxy/application_1600308539958_0616/ user: bigdata 20/09/24 03:50:41 INFO Client: Deleted staging directory hdfs://test-2:8020/user/bigdata/.sparkStaging/application_1600308539958_0616 20/09/24 03:50:41 INFO ShutdownHookManager: Shutdown hook called 20/09/24 03:50:41 INFO ShutdownHookManager: Deleting directory /tmp/spark-0d9a9713-0990-4bde-90b2-110c3e3c0d8d 20/09/24 03:50:41 INFO ShutdownHookManager: Deleting directory /tmp/spark-db51f544-ec7e-4321-aa3d-8772773aff3d

yarn集群上的状态：

FINISHED | SUCCEEDED

[xingchun-chen]

please update the title in English

[lishiyucn]

please update the title in English

Please pay attention to the question instead of the language.

[Eights-Li]

After the spark application runs, ds will use the application_id to query the application status on yarn. The application_id is stored in the app_link field of the t_ds_task_instance table. There will be a log for querying the status of the yarn application in the task instance log of ds. Please check it.

[xingchun-chen]

The title has not been updated for a long time in English, it will be closed. If there is a problem, please update the title in English and reopen it

[15177592207]

I also encountered the same problem, what should I do