hadoop-yarn-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Philip Zeyliger (JIRA)" <j...@apache.org>
Subject [jira] [Created] (YARN-1312) Job History server "queue" attribute incorrectly reports default when username is actually used for queue at runtime
Date Wed, 16 Oct 2013 02:33:42 GMT
Philip Zeyliger created YARN-1312:
-------------------------------------

             Summary: Job History server "queue" attribute incorrectly reports default when
username is actually used for queue at runtime
                 Key: YARN-1312
                 URL: https://issues.apache.org/jira/browse/YARN-1312
             Project: Hadoop YARN
          Issue Type: Bug
            Reporter: Philip Zeyliger


If you run a MapReduce job with the fair scheduler and you query the JobHistory server for
its metadata, you might see something like the following at http://jh_host:19888/ws/v1/history/mapreduce/jobs/job_1381878638171_0001/

{code}
<job>
<startTime>1381890132608</startTime>
<finishTime>1381890141988</finishTime>
<id>job_1381878638171_0001</id>
<name>TeraGen</name>
<queue>default</queue>
<user>hdfs</user>
...
</job>
{code}

The same is true if you query the RM while it's running via 
http://rm_host:8088/ws/v1/cluster/apps/application_1381878638171_0002:
{code}
<app>
<id>application_1381878638171_0002</id>
<user>hdfs</user>
<name>TeraGen</name>
<queue>default</queue>
...
</app>
{code}

As it turns out, in both of these cases, the job is actually executing in "root.hdfs" and
not in "root.default" because {{yarn.scheduler.fair.user-as-default-queue}} is set to true.

This makes it hard to figure out after the fact (or during!) what queue the MR job was running
under.



--
This message was sent by Atlassian JIRA
(v6.1#6144)

Mime
View raw message