[
https://issues.apache.org/jira/browse/YARN-2680?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14169289#comment-14169289
]
Jason Lowe commented on YARN-2680:
----------------------------------
[~djp] could you elaborate more on the use-case? During NM restart the RM is
unaware that the NM is being restarted because it's normally only down for a
few seconds. If the RM was aware of the node being down then that could easily
cause some undesired chaos (e.g.: RM informs AM of lost/removed node, AM
decides to kill container and start a new one, etc.).
> Node shouldn't be listed as RUNNING when NM daemon is stop even when recovery
> work is enabled.
> ----------------------------------------------------------------------------------------------
>
> Key: YARN-2680
> URL: https://issues.apache.org/jira/browse/YARN-2680
> Project: Hadoop YARN
> Issue Type: Bug
> Components: nodemanager
> Affects Versions: 2.5.1
> Reporter: Junping Du
> Priority: Critical
>
> After YARN-1336 (specifically saying YARN-1337), we now support container
> preserving during NM restart. During NM is down, the node shouldn't be listed
> as RUNNING from yarn node CLI or watched from RM website.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
