[ https://issues.apache.org/jira/browse/OOZIE-3061?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16357099#comment-16357099 ]
Peter Cseh commented on OOZIE-3061: ----------------------------------- Yeah, YarnClient could be more clever about this. Can you upload a cumulative patch that contains changes to both classes in question? Also, instead of checking against KILLED status I think we should only kill an application if the yarnClient.getFinalApplicationStatus() is UNDEFINED, meaning the application is still running. There is no gain in killing already succeeded/failed applications. > Kill only those child jobs which are not already killed > ------------------------------------------------------- > > Key: OOZIE-3061 > URL: https://issues.apache.org/jira/browse/OOZIE-3061 > Project: Oozie > Issue Type: Bug > Affects Versions: 4.3.0 > Reporter: Satish Subhashrao Saley > Assignee: Mate Juhasz > Priority: Trivial > Labels: newbie, newbiee > Fix For: 5.0.0 > > Attachments: OOZIE-3061-001.patch, OOZIE-3061-002.patch, > OOZIE-3061-003.patch > > > Here we kill all child jobs. > https://github.com/apache/oozie/blob/master/sharelib/oozie/src/main/java/org/apache/oozie/action/hadoop/LauncherMain.java#L265 > We should check before killing for already killed application. -- This message was sent by Atlassian JIRA (v7.6.3#76005)