You are viewing a plain text version of this content. The canonical link for it is here.
Posted to mapreduce-issues@hadoop.apache.org by "Jeffrey Naisbitt (JIRA)" <ji...@apache.org> on 2011/08/30 17:15:41 UTC

[jira] [Created] (MAPREDUCE-2914) Killed and Failed jobs are not handled correctly with NotRunningJobs and should be served by the History Server

Killed and Failed jobs are not handled correctly with NotRunningJobs and should be served by the History Server
---------------------------------------------------------------------------------------------------------------

                 Key: MAPREDUCE-2914
                 URL: https://issues.apache.org/jira/browse/MAPREDUCE-2914
             Project: Hadoop Map/Reduce
          Issue Type: Bug
          Components: jobhistoryserver
    Affects Versions: 0.23.0, 0.24.0
            Reporter: Jeffrey Naisbitt


Jobs that are that are killed or failed should be served by historyServer rather than returning the "singleton" NotRunningJob instance for that state.  Otherwise, if someone does a getJob() on a killed/failed job, he won't get the correct information at all.




--
This message is automatically generated by JIRA.
For more information on JIRA, see: http://www.atlassian.com/software/jira

[jira] [Updated] (MAPREDUCE-2914) MR jobs with the corresponding apps Killed or Failed have insufficient information in their JobStatuses

Posted by "Vinod Kumar Vavilapalli (JIRA)" <ji...@apache.org>.
     [ https://issues.apache.org/jira/browse/MAPREDUCE-2914?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Vinod Kumar Vavilapalli updated MAPREDUCE-2914:
-----------------------------------------------

    Description: 
For apps(not jobs) which are killed/failed, JobHistory is most likely garbled.

We still need to make sure that JobStatus fields like user, jobName, jobFile, trackingUrl etc are correct for jobs whose corresponding apps got killed/failed.





  was:
Jobs that are that are killed or failed should be served by historyServer rather than returning the "singleton" NotRunningJob instance for that state.  Otherwise, if someone does a getJob() on a killed/failed job, he won't get the correct information at all.




       Priority: Minor  (was: Major)
        Summary: MR jobs with the corresponding apps Killed or Failed have insufficient information in their JobStatuses  (was: Killed and Failed jobs are not handled correctly with NotRunningJobs and should be served by the History Server)

> MR jobs with the corresponding apps Killed or Failed have insufficient information in their JobStatuses
> -------------------------------------------------------------------------------------------------------
>
>                 Key: MAPREDUCE-2914
>                 URL: https://issues.apache.org/jira/browse/MAPREDUCE-2914
>             Project: Hadoop Map/Reduce
>          Issue Type: Bug
>          Components: jobhistoryserver
>    Affects Versions: 0.23.0, 0.24.0
>            Reporter: Jeffrey Naisbitt
>            Priority: Minor
>
> For apps(not jobs) which are killed/failed, JobHistory is most likely garbled.
> We still need to make sure that JobStatus fields like user, jobName, jobFile, trackingUrl etc are correct for jobs whose corresponding apps got killed/failed.

--
This message is automatically generated by JIRA.
For more information on JIRA, see: http://www.atlassian.com/software/jira

[jira] [Reopened] (MAPREDUCE-2914) Killed and Failed jobs are not handled correctly with NotRunningJobs and should be served by the History Server

Posted by "Vinod Kumar Vavilapalli (JIRA)" <ji...@apache.org>.
     [ https://issues.apache.org/jira/browse/MAPREDUCE-2914?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Vinod Kumar Vavilapalli reopened MAPREDUCE-2914:
------------------------------------------------


I also doubted this originally, but I missed the fact that for apps(not jobs) which are killed/failed, JobHistory is most likely garbled.

We still need to make sure that JobStatus fields like user, jobName, jobFile, trackingUrl etc are correct for these jobs whose corresponding apps got killed/failed.

Will update the title accordingly.

> Killed and Failed jobs are not handled correctly with NotRunningJobs and should be served by the History Server
> ---------------------------------------------------------------------------------------------------------------
>
>                 Key: MAPREDUCE-2914
>                 URL: https://issues.apache.org/jira/browse/MAPREDUCE-2914
>             Project: Hadoop Map/Reduce
>          Issue Type: Bug
>          Components: jobhistoryserver
>    Affects Versions: 0.23.0, 0.24.0
>            Reporter: Jeffrey Naisbitt
>
> Jobs that are that are killed or failed should be served by historyServer rather than returning the "singleton" NotRunningJob instance for that state.  Otherwise, if someone does a getJob() on a killed/failed job, he won't get the correct information at all.

--
This message is automatically generated by JIRA.
For more information on JIRA, see: http://www.atlassian.com/software/jira

[jira] [Resolved] (MAPREDUCE-2914) Killed and Failed jobs are not handled correctly with NotRunningJobs and should be served by the History Server

Posted by "Sharad Agarwal (JIRA)" <ji...@apache.org>.
     [ https://issues.apache.org/jira/browse/MAPREDUCE-2914?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Sharad Agarwal resolved MAPREDUCE-2914.
---------------------------------------

    Resolution: Invalid

Application killed/failed is different from Job killed/failed. If the job is killed/failed in AM, Application state would still be successful. Successful application means AM exited gracefully after doing the unregister.
If AM is crashed or forcefully killed, in that case history file is not consistent. So history server can't serve the requests.

Please feel free to reopen if you disagree.

> Killed and Failed jobs are not handled correctly with NotRunningJobs and should be served by the History Server
> ---------------------------------------------------------------------------------------------------------------
>
>                 Key: MAPREDUCE-2914
>                 URL: https://issues.apache.org/jira/browse/MAPREDUCE-2914
>             Project: Hadoop Map/Reduce
>          Issue Type: Bug
>          Components: jobhistoryserver
>    Affects Versions: 0.23.0, 0.24.0
>            Reporter: Jeffrey Naisbitt
>
> Jobs that are that are killed or failed should be served by historyServer rather than returning the "singleton" NotRunningJob instance for that state.  Otherwise, if someone does a getJob() on a killed/failed job, he won't get the correct information at all.

--
This message is automatically generated by JIRA.
For more information on JIRA, see: http://www.atlassian.com/software/jira