[ https://issues.apache.org/jira/browse/YARN-7225?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16670827#comment-16670827 ]
Eric Payne commented on YARN-7225: ---------------------------------- Thanks [~jhung]. The behavior is slightly different depending on if we are using {{appAMNodePartitionName}} vs. { {{partition}}, {{schedulerContainer.getSchedulerNode().getPartition()}}, {{node.getPartition()}} } 1) {{appAMNodePartitionName}} will always be the partition of the application. 2) Using the other methods to get the partition AND non-exclusive labels is used, it will be the label of the node on which the container ran. I think #2 is technically the correct behavior, but I also think that it may be confusing if the application was submitted to a queue that does not explicitly allow that label. Thoughts? > Add queue and partition info to RM audit log > -------------------------------------------- > > Key: YARN-7225 > URL: https://issues.apache.org/jira/browse/YARN-7225 > Project: Hadoop YARN > Issue Type: Improvement > Components: resourcemanager > Affects Versions: 2.9.1, 2.8.4, 3.0.2, 3.1.1 > Reporter: Jonathan Hung > Assignee: Eric Payne > Priority: Major > Attachments: YARN-7225.001.patch, YARN-7225.002.patch, > YARN-7225.003.patch, YARN-7225.004.patch, YARN-7225.005.patch, > YARN-7225.branch-2.8.001.patch > > > Right now RM audit log has fields such as user, ip, resource, etc. Having > queue and partition is useful for resource tracking. -- 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