[ https://issues.apache.org/jira/browse/YARN-2802?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14213807#comment-14213807 ]
Vinod Kumar Vavilapalli commented on YARN-2802: ----------------------------------------------- QueueMetrics are modeled around a limited number of queues each with limited information related to the queue. IIUC, what you really need is some sort of Cluster metrics - the information you are adding is not so useful per queue, correct? If so, we can put it into org.apache.hadoop.yarn.server.resourcemanager.ClusterMetrics? > add AM container launch and register delay metrics in QueueMetrics to help > diagnose performance issue. > ------------------------------------------------------------------------------------------------------ > > Key: YARN-2802 > URL: https://issues.apache.org/jira/browse/YARN-2802 > Project: Hadoop YARN > Issue Type: Improvement > Components: resourcemanager > Affects Versions: 2.5.0 > Reporter: zhihai xu > Assignee: zhihai xu > Attachments: YARN-2802.000.patch, YARN-2802.001.patch, > YARN-2802.002.patch > > > add AM container launch and register delay metrics in QueueMetrics to help > diagnose performance issue. > Added two metrics in QueueMetrics: > aMLaunchDelay: the time spent from sending event AMLauncherEventType.LAUNCH > to receiving event RMAppAttemptEventType.LAUNCHED in RMAppAttemptImpl. > aMRegisterDelay: the time waiting from receiving event > RMAppAttemptEventType.LAUNCHED to receiving event > RMAppAttemptEventType.REGISTERED(ApplicationMasterService#registerApplicationMaster) > in RMAppAttemptImpl. -- This message was sent by Atlassian JIRA (v6.3.4#6332)