hadoop-mapreduce-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Jason Lowe (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (MAPREDUCE-5086) MR app master deletes staging dir when sent a reboot command from the RM
Date Thu, 28 Mar 2013 18:25:16 GMT

    [ https://issues.apache.org/jira/browse/MAPREDUCE-5086?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13616504#comment-13616504
] 

Jason Lowe commented on MAPREDUCE-5086:
---------------------------------------

bq. I am fine with changing the interface except for the fact that it exposes the state machine
concept that is internal to JobImpl.

That's sort of my point with casting to JobImpl, it effectively throws away the fact that
we have an interface and goes straight to the implementation.  I understand the motiviation,
just lamenting that there should be a mechanism via the interface to accomplish the same semantics.
 We don't have to expose the entire state machine, instead wondering if it makes sense to
have something like a simple boolean interface that encapsulates the concept of what we're
checking for during shutdown.

Bottom line is I won't block this going in if the cast remains, but if it does there should
be a followup JIRA to try to address it.
                
> MR app master deletes staging dir when sent a reboot command from the RM
> ------------------------------------------------------------------------
>
>                 Key: MAPREDUCE-5086
>                 URL: https://issues.apache.org/jira/browse/MAPREDUCE-5086
>             Project: Hadoop Map/Reduce
>          Issue Type: Bug
>            Reporter: jian he
>            Assignee: jian he
>         Attachments: YARN-472.1.patch, YARN-472.2.patch, YARN-472.3.patch, YARN-472.4.patch,
YARN-472.5.patch, YARN-472.6.patch
>
>
> If the RM is restarted when the MR job is running, then it sends a reboot command to
the job. The job ends up deleting the staging dir and that causes the next attempt to fail.

--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira

Mime
View raw message