[ https://issues.apache.org/jira/browse/YARN-2889?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16815648#comment-16815648 ]
Íñigo Goiri commented on YARN-2889: ----------------------------------- Thanks [~abmodi] for the patch. A few comments: * Can we use "opportunistic containers" in {{yarn-default.xml}}? * The main logic seems to be in {{OpportunisticContainerAllocator}}, it could use some logging (both info and debug) to understand why stuff is getting capped. * In {{TestOpportunisticContainerAllocator}}, we could extract {{ExecutionTypeRequest.newInstance(ExecutionType.OPPORTUNISTIC, true))}} as well as {{Priority.newInstance(1)}} into a constant. * The tests in {{TestOpportunisticContainerAllocator}} could use some comments and probably even a javadoc explaining what is expected. * In {{TestOpportunisticContainerAllocator}}, the System.out should be a LOG (together with the comments). > Limit the number of opportunistic container allocated per AM heartbeat > ---------------------------------------------------------------------- > > Key: YARN-2889 > URL: https://issues.apache.org/jira/browse/YARN-2889 > Project: Hadoop YARN > Issue Type: Sub-task > Components: nodemanager, resourcemanager > Reporter: Konstantinos Karanasos > Assignee: Abhishek Modi > Priority: Major > Attachments: YARN-2889.001.patch, YARN-2889.002.patch > > > We introduce a way to limit the number of opportunistic containers that will > be allocated on each AM heartbeat. > This way we can restrict the number of opportunistic containers handed out > by the system, as well as throttle down misbehaving AMs (asking for too many > opportunistic containers). -- This message was sent by Atlassian JIRA (v7.6.3#76005) --------------------------------------------------------------------- To unsubscribe, e-mail: yarn-issues-unsubscr...@hadoop.apache.org For additional commands, e-mail: yarn-issues-h...@hadoop.apache.org