[
https://issues.apache.org/jira/browse/HIVE-9774?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Rui Li updated HIVE-9774:
-------------------------
Attachment: HIVE-9774.1-spark.patch
The patch uses {{SparkContext::applicationId}}, which is the YARN app ID when
spark runs on yarn.
It's printing "Running with YARN application" instead of "Starting
application". This is because user can submit multiple jobs to one spark
session -- we're not starting an application for each job.
> Print yarn application id to console [Spark Branch]
> ---------------------------------------------------
>
> Key: HIVE-9774
> URL: https://issues.apache.org/jira/browse/HIVE-9774
> Project: Hive
> Issue Type: Sub-task
> Components: Spark
> Reporter: Brock Noland
> Assignee: Rui Li
> Attachments: HIVE-9774.1-spark.patch
>
>
> Oozie would like to use beeline to capture the yarn application id of apps so
> that if a workflow is canceled, the job can be cancelled. When running under
> MR we print the job id but under spark we do not.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
