[ https://issues.apache.org/jira/browse/YARN-3328?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15073587#comment-15073587 ]
sandflee commented on YARN-3328: -------------------------------- we fix this problem by removing container state machine in NMClientAsyncImpl, and has worked for half a year. Is this solution acceptable? > There's no way to rebuild containers Managed by NMClientAsync If AM restart > --------------------------------------------------------------------------- > > Key: YARN-3328 > URL: https://issues.apache.org/jira/browse/YARN-3328 > Project: Hadoop YARN > Issue Type: Bug > Components: api, applications, client > Affects Versions: 2.6.0 > Reporter: sandflee > > If work preserving is enabled and AM restart, AM could't stop containers > launched by pre-am, because there's no corresponding container in > NMClientAsync.containers. > Thereās no way to rebuild NMClientAsync.containers. -- This message was sent by Atlassian JIRA (v6.3.4#6332)