[ https://issues.apache.org/jira/browse/YARN-41?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14548130#comment-14548130 ]
Jason Lowe commented on YARN-41: -------------------------------- Looking at the table above, I'm wondering about the case where we are doing a graceful shutdown, recovery is enabled, but we are not running under supervision. When we are shutting down without supervision the NM will normally kill active containers, so in that sense I think we should also unregister. I'm not sure there's a point to avoiding the unregister if no containers will survive the NM shutdown. The NM only avoids killing containers on shutdown if the NM supports recovery and it has been told it is being supervised (i.e.: it is likely the NM will be restarted shortly after the shutdown completes). In the other cases it kills containers on shutdown to avoid a situation where containers are running uncontrolled on a node due to the NM being unavailable for a prolonged duration. > The RM should handle the graceful shutdown of the NM. > ----------------------------------------------------- > > Key: YARN-41 > URL: https://issues.apache.org/jira/browse/YARN-41 > Project: Hadoop YARN > Issue Type: New Feature > Components: nodemanager, resourcemanager > Reporter: Ravi Teja Ch N V > Assignee: Devaraj K > Labels: BB2015-05-TBR > Attachments: MAPREDUCE-3494.1.patch, MAPREDUCE-3494.2.patch, > MAPREDUCE-3494.patch, YARN-41-1.patch, YARN-41-2.patch, YARN-41-3.patch, > YARN-41-4.patch, YARN-41.patch > > > Instead of waiting for the NM expiry, RM should remove and handle the NM, > which is shutdown gracefully. -- This message was sent by Atlassian JIRA (v6.3.4#6332)