[ https://issues.apache.org/jira/browse/YARN-2488?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
Xu Yang updated YARN-2488: -------------------------- Issue Type: Improvement (was: Bug) > Nodemanager should support check if bad disks have turned to good ones > ---------------------------------------------------------------------- > > Key: YARN-2488 > URL: https://issues.apache.org/jira/browse/YARN-2488 > Project: Hadoop YARN > Issue Type: Improvement > Components: nodemanager > Reporter: Xu Yang > > When NM's disks turn bad, it will never turn good ones. Although the disks is > fixed, we must restart nodemanager to use the fixed disks. In particular when > NM become UNHEALTHY status, it will never be back to HEALTHY status. -- This message was sent by Atlassian JIRA (v6.3.4#6332)