[jira] [Comment Edited] (HBASE-17601) close() in TableRecordReaderImpl assumes the split has started

["Patrick Dignan (JIRA)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/HBASE-17601?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15854543#comment-15854543 ] 

Patrick Dignan edited comment on HBASE-17601 at 2/6/17 6:57 PM:
----------------------------------------------------------------

I'm mostly unfamiliar with the HBase patching/branching workflow, but it looks like this patch isn't merged into branch-1.{0,1,2,3}. Is there something else that needs to happen for it to be merged into those, or will that happen later as part of a testing process?


was (Author: pdignan):
I'm mostly unfamiliar with the HBase patching/branching workflow, but it looks like this patch isn't merged into branch-1.{0,1,2,3}.  Is there something else that needs to happen for it to be merged into those, or will that happen later as part of a testing process?

> close() in TableRecordReaderImpl assumes the split has started
> --------------------------------------------------------------
>
>                 Key: HBASE-17601
>                 URL: https://issues.apache.org/jira/browse/HBASE-17601
>             Project: HBase
>          Issue Type: Bug
>          Components: hadoop2
>            Reporter: Michael Axiak
>            Assignee: Michael Axiak
>            Priority: Minor
>             Fix For: 2.0.0, 1.4.0
>
>         Attachments: HBASE-17601.patch
>
>
> When closing the input format, it assumes that the split has already started, which means that close can cause exceptions. In the case the map task decides not to fetch a single row, this causes the job to unnecessarily fail.



--
This message was sent by Atlassian JIRA
(v6.3.15#6346)