[jira] [Commented] (MAPREDUCE-3418) If map output is not found, shuffle runs in tight loop

["archit neema (JIRA)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/MAPREDUCE-3418?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15050915#comment-15050915 ] 

archit neema commented on MAPREDUCE-3418:
-----------------------------------------

This issue replicated at my end 
Reason :- 
previously hadoop was running in unsecured (kerberos) environment
And i ran Maprjobs
appcache Directory was created inside the yarn//nm/usercache/username/ 
Automatically access rights on this directory is drwx--x---

currently i switch my hadoop from unsure to secure environment
i am not able to run any mapreduce jobs 

Workaround :
Go To yarn//nm/usercache/username/
Delete appcache Directory, Restart yarn service 
and re-run the failure jobs, Automatically appcache will created with  drwx--S--- Access rights 

Possible Solution:
while switching from unsecured environment to  secured Environment then Access rights should be check on all the directories  




> If map output is not found, shuffle runs in tight loop
> ------------------------------------------------------
>
>                 Key: MAPREDUCE-3418
>                 URL: https://issues.apache.org/jira/browse/MAPREDUCE-3418
>             Project: Hadoop Map/Reduce
>          Issue Type: Bug
>    Affects Versions: 0.23.0, 2.3.0
>            Reporter: John George
>            Assignee: sahitya pavurala
>
> Sharad Agarwal bumped into this while simulating fetch failures. 
> Removed the map output directory. Shuffle runs in tight loop throwing
> :
> 2011-06-01 09:02:20,511 WARN org.apache.hadoop.mapreduce.task.reduce.Fetcher: Invalid map id 
> java.lang.IllegalArgumentException: TaskAttemptId string : TTP/1.1 500 Internal Server Error
> Content-Type: text/plain; charset=UTF is not properly formed
>     at org.apache.hadoop.mapreduce.TaskAttemptID.forName(TaskAttemptID.java:174)
>     at org.apache.hadoop.mapreduce.task.reduce.Fetcher.copyMapOutput(Fetcher.java:284)
>     at org.apache.hadoop.mapreduce.task.reduce.Fetcher.copyFromHost(Fetcher.java:251)
>     at org.apache.hadoop.mapreduce.task.reduce.Fetcher.run(Fetcher.java:149)
> Fetch failure is not triggered.



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