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 "Neelesh Srinivas Salian (JIRA)" <ji...@apache.org> on 2015/09/24 17:41:05 UTC

[jira] [Updated] (MAPREDUCE-6449) MR Code should not throw and catch YarnRuntimeException to communicate internal exceptions

     [ https://issues.apache.org/jira/browse/MAPREDUCE-6449?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Neelesh Srinivas Salian updated MAPREDUCE-6449:
-----------------------------------------------
    Attachment: MAPREDUCE-6449.001.patch

Attached the 1st version of this patch. 
1) I created MRRuntimeException.java to be equivalent to YarnRuntimeException and extends RuntimeException to help being replaced in the catch and throw blocks.

2) Replaced the calls to YarnRuntimeException in most of the files.

Concerns and questions:
1) Would the v2/app/exceptions be the right location for such a file or should it be in hadoop-mapreduce-client-common to be better accessible?
2) I went ahead and pointed all the YarnRuntimeException to MRRuntimeException. Should this cover all the places including the ones that catch(IOException e) { throws new YarnRuntimeException (e)}

Seen this in multiple instances, and I was not too sure about modifying this.

Requesting review on this version.
I will change/augment based on the review and add tests accordingly.
Planning to have TestSuite.java for exceptions.

Thank you.

> MR Code should not throw and catch YarnRuntimeException to communicate internal exceptions
> ------------------------------------------------------------------------------------------
>
>                 Key: MAPREDUCE-6449
>                 URL: https://issues.apache.org/jira/browse/MAPREDUCE-6449
>             Project: Hadoop Map/Reduce
>          Issue Type: Bug
>            Reporter: Anubhav Dhoot
>            Assignee: Neelesh Srinivas Salian
>         Attachments: MAPREDUCE-6449.001.patch
>
>
> In discussion of MAPREDUCE-6439 we discussed how throwing and catching YarnRuntimeException in MR code is incorrect and we should instead use some MR specific exception.



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