hadoop-yarn-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Carlo Curino (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (YARN-5185) StageAllocaterGreedyRLE: NPE in corner case
Date Mon, 06 Jun 2016 21:40:21 GMT

    [ https://issues.apache.org/jira/browse/YARN-5185?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15317311#comment-15317311
] 

Carlo Curino commented on YARN-5185:
------------------------------------

[~asuresh], this bug is very similar to what we had in YARN-4525, for which I was able to
construct a test. Here it is tricky, as the bug only triggers once the NavigableMap is accessed
and restricted multiple times, and the use is fairly deep into the stack of calls we can test.


However (answering your question), I saw tests in YARN-5164 triggering this bug. I am ok merging
this or not as you prefer. I think the patch is fairly clearly non-harmful and protects against
a possible NPE so I would suggest to commit it as is, because the reviewing of YARN-5164 
is less trivial, and actually quite delicate.

> StageAllocaterGreedyRLE: NPE in corner case 
> --------------------------------------------
>
>                 Key: YARN-5185
>                 URL: https://issues.apache.org/jira/browse/YARN-5185
>             Project: Hadoop YARN
>          Issue Type: Sub-task
>          Components: capacityscheduler, fairscheduler, resourcemanager
>            Reporter: Carlo Curino
>            Assignee: Carlo Curino
>         Attachments: YARN-5185.1.patch
>
>
> If the plan has only one interval, and the reservation exactly overlap we will have a
null from partialMap.higherKey() that we should guard against.



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

---------------------------------------------------------------------
To unsubscribe, e-mail: yarn-issues-unsubscribe@hadoop.apache.org
For additional commands, e-mail: yarn-issues-help@hadoop.apache.org


Mime
View raw message