[ 
https://issues.apache.org/jira/browse/MAPREDUCE-3098?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13116417#comment-13116417
 ] 

Vinod Kumar Vavilapalli commented on MAPREDUCE-3098:
----------------------------------------------------

I understand the areas of concern now. To summarize, let's do this:
 - Rename ApplicationState to _YarnApplicationState_ without changing any 
functionality.
 - Rename ApplicationState.SUCCEEDED to _YarnApplicationState.FINISHED_
 - Make "String finalState" as "Enum _FinalStatus_" which can take values 
SUCCEEDED, FAILED and KILLED.

Now the various scenarios:
 - If an application crashes because of Yarn issues, for e.g. AM going over 
limits, _YarnApplicationState_ will be FAILED.
 - If an application fails because of framework issue, for e.g a MR task fails 
four times and thus the job fails, _YarnApplicationState_ will be FINISHED 
reflecting no platform errors, but AM will send _FinalStatus_ as FAILED
 - If an application finishes properly and the framework is happy too, for e.g. 
a successful MR job, _YarnApplicationState_ will be FINISHED and AM will send 
_FinalStatus_ to be SUCCEEDED.

To summarize, for the client, the following holds:
||_YarnApplicationState_||_FinalStatus_||What Happenend||
|FINISHED|SUCCEEDED|MR job succeeeded|
|FINISHED|FAILED|MR job failed say due to four attempt failures|
|FINISHED|KILLED|MR job killed by user via AM|
|FAILED|Anything|MR job failed because of YARN errors|
|KILLED|Anything|MR job forcefully terminated via RM|

We will need separate columns on the RM UI for _YarnApplicationState_ and 
_FinalStatus_. Similarly, we also need separate columns on the command line for 
"bin/hadoop job -list" and an extra entry for _YarnApplicationState_ in 
"bin/hadoop job -status"
                
> Report Application status as well as ApplicationMaster status in 
> GetApplicationReportResponse 
> ----------------------------------------------------------------------------------------------
>
>                 Key: MAPREDUCE-3098
>                 URL: https://issues.apache.org/jira/browse/MAPREDUCE-3098
>             Project: Hadoop Map/Reduce
>          Issue Type: Sub-task
>          Components: mrv2
>    Affects Versions: 0.23.0
>            Reporter: Hitesh Shah
>            Assignee: Hitesh Shah
>            Priority: Blocker
>             Fix For: 0.23.0
>
>
> Currently, an application report received by the client from the RM/ASM for a 
> given application returns the status of the application master. It does not 
> return the status of the application i.e. whether that particular job 
> succeeded or failed. 
> The AM status would be one of FINISHED (SUCCEEDED should be renamed to 
> FINISHED as AM state does not indicate overall success/failure), FAILED or 
> KILLED. 
> The final state sent by the AM to the RM in the 
> FinishApplicationMasterRequest should be exposed to the client as 
> ApplicationState. 

--
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

        

Reply via email to