[ https://issues.apache.org/jira/browse/YARN-4051?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14996345#comment-14996345 ]
sandflee commented on YARN-4051: -------------------------------- Is it possible for the finish application or complete container requests to arrive at this point? yes, we see this in YARN-4050. If we register to RM after complete container recover, we must face the risk that the container running on this node will be killed if container recovery takes much more time(in YARN-4050), for long-runing-services, maybe not so perfect. > ContainerKillEvent is lost when container is In New State and is recovering > ---------------------------------------------------------------------------- > > Key: YARN-4051 > URL: https://issues.apache.org/jira/browse/YARN-4051 > Project: Hadoop YARN > Issue Type: Bug > Components: nodemanager > Reporter: sandflee > Assignee: sandflee > Priority: Critical > Attachments: YARN-4051.01.patch, YARN-4051.02.patch, > YARN-4051.03.patch > > > As in YARN-4050, NM event dispatcher is blocked, and container is in New > state, when we finish application, the container still alive even after NM > event dispatcher is unblocked. -- This message was sent by Atlassian JIRA (v6.3.4#6332)