[ https://issues.apache.org/jira/browse/MAPREDUCE-6449?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14971033#comment-14971033 ]
Junping Du commented on MAPREDUCE-6449: --------------------------------------- Is this patch breaking our MR rolling upgrade story that old MR job and new MR job can coexist in a single cluster (during upgrade)? At least, changes on hs part sounds like this. If so, I would be very concern on this. > 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 > Labels: mapreduce > Attachments: MAPREDUCE-6449.001.patch, MAPREDUCE-6449.002.patch, > MAPREDUCE-6499-prelim.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)