[ 
https://issues.apache.org/jira/browse/KNOX-715?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15306993#comment-15306993
 ] 

Pierre Regazzoni commented on KNOX-715:
---------------------------------------

[~lmccay] That's correct it will generate files in input_terasort directory. I 
think 60s might be too close to wait for the job to finish potentially. I just 
ran example on my cluster and tweak the parameters a bit (500 records with 5 
mappers) to be more conservative:

{code}
Delete /user/biadmin/test: 200
Create /user/biadmin/test: 200
Put /user/biadmin/test/lib/hadoop-mapreduce-examples.jar: 201
Submitted job: job_1461087894116_1157
Polling up to 60s for job completion...
...............................
Job status: true
[_SUCCESS, part-m-00000, part-m-00001, part-m-00002, part-m-00003, part-m-00004]
Session closed: true
{code}

Can you check the actual job log or try to change the param as follow:

{code}
  .arg( "-D").arg("mapred.map.tasks=5") \
  .arg( "500" ) \
{code}

Let me know if that works so I can deliver a new version. I will also add some 
instruction (as comment) in the sample regarding getting the jar per your 
suggestion (e.g. 
/usr/[iop|hdp]/current/hadoop-mapreduce-client/hadoop-mapreduce-examples.jar)

> submitJava should allow multiple argument via arg()
> ---------------------------------------------------
>
>                 Key: KNOX-715
>                 URL: https://issues.apache.org/jira/browse/KNOX-715
>             Project: Apache Knox
>          Issue Type: Bug
>          Components: ClientDSL
>    Affects Versions: 0.7.0
>            Reporter: Pierre Regazzoni
>            Assignee: Pierre Regazzoni
>             Fix For: 0.9.1
>
>         Attachments: KNOX-715-002.patch, KNOX-715.patch
>
>
> Like with submitPig (KNOX-712). I want to be able to provide multiple 
> argument when submitting a jar via submitJava. For example, should be able to 
> do:
> {noformat}
> jobId = Job.submitJava(session) \
>      .jar( "path-to-my-jar" ) \
>      .app( "class-name" ) \
>      .arg( "argument1").arg("value1") \
>      .arg( "argument2").arg("value2") \
>      .arg( "argument3") \
>      .output("path-to-output-dir ) \
>      .now().jobId
> {noformat}
> Current code limits execution to only specifying jar,class,input,output. I 
> should have a patch for it.



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)

Reply via email to