[jira] [Commented] (HBASE-11369) The proceduce of interrupting current split task should be updated after HBASE-9736.

["Jeffrey Zhong (JIRA)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/HBASE-11369?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14034037#comment-14034037 ] 

Jeffrey Zhong commented on HBASE-11369:
---------------------------------------

Yes, the interrupted logic should be updated as well. Since updates need to verify znode version, the task will fail at the end. Will you want to come a patch for this? Thanks.

> The proceduce of interrupting current split task should be updated after HBASE-9736.
> ------------------------------------------------------------------------------------
>
>                 Key: HBASE-11369
>                 URL: https://issues.apache.org/jira/browse/HBASE-11369
>             Project: HBase
>          Issue Type: Bug
>          Components: wal
>            Reporter: Liu Shaohui
>            Priority: Minor
>
> Before HBASE-9736, SplitLogWorker only split one hlog at a time. When the data of znode for this task is changed (task is timeouted and resigned by splitLogManager), zookeeper will notify the SplitLogWorker. If this log task is owned by other regionserver, the SplitLogWorker will interrupt current task and try to get another task.
> HBASE-9736 allow multi log splitters per RS so there will be multi current tasks running in the thread pool in SplitLogWorker.
> So the proceduce of Interrupting current split task need be updated.
> [~jeffreyz]  [~stack]



--
This message was sent by Atlassian JIRA
(v6.2#6252)