如何从外壳脚本中捕获Spark错误

Question

我在 AWS 数据管道中有一个管道，该管道运行一个名为 shell.sh 的 shell 脚本：

$ spark-submit transform_json.py

Running command on cluster...
[54.144.10.162] Running command...
[52.206.87.30] Running command...
[54.144.10.162] Command complete.
[52.206.87.30] Command complete.
run_command finished in 0:00:06.

AWS 数据管道控制台显示作业"已完成"，但在 stderr 日志中，我看到作业实际上已中止：

Caused by: com.amazonaws.services.s3.model.AmazonS3Exception: Status Code: 404, AWS Service: Amazon S3, AWS Request ID: xxxxx, AWS Error Code: null, AWS Error Message: Not Found...        
Caused by: org.apache.spark.SparkException: Job aborted due to stage failure: Task 0 in stage 5.0 failed 1 times, most recent failure: Lost task 0.0 in stage 5.0 (TID 5, localhost, executor driver): org.apache.spark.SparkException: Task failed while writing rows.
...
20/05/22 11:42:47 INFO MapOutputTrackerMasterEndpoint: MapOutputTrackerMasterEndpoint stopped!
20/05/22 11:42:47 INFO MemoryStore: MemoryStore cleared
20/05/22 11:42:47 INFO BlockManager: BlockManager stopped
20/05/22 11:42:47 INFO BlockManagerMaster: BlockManagerMaster stopped
20/05/22 11:42:47 INFO OutputCommitCoordinator$OutputCommitCoordinatorEndpoint: OutputCommitCoordinator stopped!
20/05/22 11:42:47 INFO SparkContext: Successfully stopped SparkContext
20/05/22 11:42:47 INFO ShutdownHookManager: Shutdown hook called

我对数据管道和Spark有点陌生;无法理解幕后实际发生的事情。如何让 shell 脚本捕获SparkException？

Answer 1

尝试下面的例子...

您的 shell 脚本可以捕获这样的错误代码...其中非零退出代码为错误

$?是最近执行的命令的退出状态;按照惯例，0 表示成功，其他任何值表示失败。

spark-submit transform_json.py

ret_code=$?
if [ $ret_code -ne 0 ]; then 
exit $ret_code
fi

您必须编写代码才能在错误条件下sys.exit(-1)返回退出代码。 检查这个 Python 异常处理...

在 Python 中检查此退出代码