[
https://issues.apache.org/jira/browse/HADOOP-14995?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16256378#comment-16256378
]
DENG FEI commented on HADOOP-14995:
-----------------------------------
met the problem again, and upload more attachments.
In mapreduce 2.2.0, the yarnchild will load "job.xml" twice, and the last one
is configured for task.
Through analyze the log, we found the second times job.xml is missing.
> Configuration should load resource by forced as option
> ------------------------------------------------------
>
> Key: HADOOP-14995
> URL: https://issues.apache.org/jira/browse/HADOOP-14995
> Project: Hadoop Common
> Issue Type: Improvement
> Components: conf
> Reporter: DENG FEI
> Priority: Critical
> Attachments: HADOOP-14995.trunk-001.patch,
> HADOOP-14995.trunk-2.patch, data_loss.png, job_detail.png, log.png,
> normal_task.png
>
>
> MapReduce Job, the task initialization by _job.xml_ , but if missing the soft
> link of "job.xml" file,will do nothing under the quietmode of Configuration.
> Such as map tasks as directly output to HDFS without reduce tasks of job, if
> map task missing job.xml, it's will run with default behavior, just as
> _IdentityMapper_ and output to local, and resulting in data loss.
--
This message was sent by Atlassian JIRA
(v6.4.14#64029)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
