[ https://issues.apache.org/jira/browse/YARN-596?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
Wei Yan updated YARN-596: ------------------------- Attachment: YARN-596.patch [~sandyr], [~kkambatl]. Could you guys take a look of this patch? The premption rules right now. For FSQueue: select the child candidate queue/app in reverse of its scheduling policy (fair, drf, or fifo). For AppScheduable: select the candidate container in decreasing order of priority. And I moved all preemption-related code from FairScheduler.java to a separate file FSPreemption.java. > In fair scheduler, intra-application container priorities affect > inter-application preemption decisions > ------------------------------------------------------------------------------------------------------- > > Key: YARN-596 > URL: https://issues.apache.org/jira/browse/YARN-596 > Project: Hadoop YARN > Issue Type: Bug > Components: scheduler > Affects Versions: 2.0.3-alpha > Reporter: Sandy Ryza > Assignee: Sandy Ryza > Attachments: YARN-596.patch > > > In the fair scheduler, containers are chosen for preemption in the following > way: > All containers for all apps that are in queues that are over their fair share > are put in a list. > The list is sorted in order of the priority that the container was requested > in. > This means that an application can shield itself from preemption by > requesting it's containers at higher priorities, which doesn't really make > sense. > Also, an application that is not over its fair share, but that is in a queue > that is over it's fair share is just as likely to have containers preempted > as an application that is over its fair share. -- This message was sent by Atlassian JIRA (v6.2#6252)