[ 
https://issues.apache.org/jira/browse/OOZIE-1052?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13505885#comment-13505885
 ] 

Ryota Egashira commented on OOZIE-1052:
---------------------------------------

did additional test where I commented out following part in 
LauncherMapper.hasIdSwap and ran workflow with map-reduce action, which worked. 
====
        if (counters != null) {
            Counters.Group counterGroup = counters.getGroup(COUNTER_GROUP);
            if (counterGroup != null) {
                swap = counterGroup.getCounter(COUNTER_DO_ID_SWAP) == 1;
        }
=====

                
> HadoopAccessorService.createFileSystem throws exception in map-reduce action, 
> failing workflow.
> -----------------------------------------------------------------------------------------------
>
>                 Key: OOZIE-1052
>                 URL: https://issues.apache.org/jira/browse/OOZIE-1052
>             Project: Oozie
>          Issue Type: Bug
>          Components: core
>    Affects Versions: 3.3.0, trunk
>            Reporter: Ryota Egashira
>            Assignee: Ryota Egashira
>            Priority: Blocker
>             Fix For: 3.3.0, trunk
>
>         Attachments: OOZIE-1052.patch
>
>
> HadoopAccessorService.createFileSystem checks 
> OOZIE_HADOOP_ACCESSOR_SERVICE_CREATED in jobconf, (which is passed as 
> argument), and if false or non-exist, throws exception. 
> ------
> if (!conf.getBoolean(OOZIE_HADOOP_ACCESSOR_SERVICE_CREATED, false)) {
>             throw new HadoopAccessorException(ErrorCode.E0903);
> }
> ------
> this is troublesome when it's called from LauncherMapper.hasIdSwap, since it 
> passes new conf as argument, which doesn't include 
> OOZIE_HADOOP_ACCESSOR_SERVICE_CREATED, so always hitting exception mentioned 
> above 
> -------
> FileSystem fs = 
> Services.get().get(HadoopAccessorService.class).createFileSystem(user, 
> p.toUri(),new Configuration());
> --------
> this problem becomes visible to user as workflow failure, when in map-reduce 
> action, oozie fails to get external jobID from counters, then accessing to 
> HDFS to get it.
> the above-mentioned change in HadoopAccessorService.createFileSystem is 
> introduced in 3.2,  it is not present in 3.1.4.
> this needs to be fixed. also question is what was original intention of the 
> change. 

--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira

Reply via email to