[
https://issues.apache.org/jira/browse/MAPREDUCE-4080?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13241231#comment-13241231
]
Bhallamudi Venkata Siva Kamesh commented on MAPREDUCE-4080:
-----------------------------------------------------------
This is because of the *stale* object returned from the *loadedJobCache*.
When I tried to access the job config first time, when job history files were
present in the intermediate directory, as *loadedJobCache* does not contain any
entries to fulfill the request, CompletedJob object was created and has been
added to *loadedJobCache* and the same object has been returned. However when I
tried to access the same job config again, this time job history files were
present in the done directory, as *loadedJobCache* already contained the job
entry, *loadedJobCache* returned the job object from its cache. So the returned
job object's confFile path pointing to *intermediate* directory where as the
job's actual confFile is present inside done directory and hence the exception.
> FileNotFoundException while accessing job configuration from UI.
> ----------------------------------------------------------------
>
> Key: MAPREDUCE-4080
> URL: https://issues.apache.org/jira/browse/MAPREDUCE-4080
> Project: Hadoop Map/Reduce
> Issue Type: Bug
> Components: jobhistoryserver, mrv2
> Affects Versions: 2.0.0, trunk
> Reporter: Bhallamudi Venkata Siva Kamesh
> Priority: Critical
>
> Tried to access the job configuration from UI, when the job history files
> were still in the intermediate directory. JHS displayed the configurations of
> the job. Again tried to access the configurations of the same job, when the
> job history files were in the done directory. This time got the following
> exception
> {noformat}
> java.io.FileNotFoundException: File does not exist:
> /jobhistory/intermediate/kamesh/job_1332999698561_0005_conf.xml
> at
> org.apache.hadoop.hdfs.server.namenode.FSNamesystem.getBlockLocationsUpdateTimes(FSNamesystem.java:1153)
> at
> org.apache.hadoop.hdfs.server.namenode.FSNamesystem.getBlockLocations(FSNamesystem.java:1115)
> at
> org.apache.hadoop.hdfs.server.namenode.FSNamesystem.getBlockLocations(FSNamesystem.java:1087)
> {noformat}
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators:
https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa
For more information on JIRA, see: http://www.atlassian.com/software/jira
