[ https://issues.apache.org/jira/browse/YARN-4712?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15184400#comment-15184400 ]
Naganarasimha G R commented on YARN-4712: ----------------------------------------- Thanks [~sjlee0] for the clarification with example and agree that there is no point in summing up {{cpuUsageTotalCoresPercentage}} but there are issues with other approach( aggregating *cpuUsagePercentPerCore*) which you have also mentioned # total number of cores for the cluster would be required to actually arrive at a percentage and also it would involve complexities when nodes go down intermittently before the aggregation and effective usage should not be more than the actual cluster cores # total number of cores should not be just the machines cores but actually which is made available to YARN (*nodeCpuPercentageForYARN*) # Still it might not be better comparison as the type of the core might also be different in a heterogenous cluster. Usually we try to have some multiplication factor so that vcores overall match. so may be we need to consider this factor too ? But anyway this would require more discussion on what would be the right one to choose so i will raise a new jira as this bug we have currently will block some one testing the ATSv2 > CPU Usage Metric is not captured properly in YARN-2928 > ------------------------------------------------------ > > Key: YARN-4712 > URL: https://issues.apache.org/jira/browse/YARN-4712 > Project: Hadoop YARN > Issue Type: Sub-task > Components: timelineserver > Reporter: Naganarasimha G R > Assignee: Naganarasimha G R > Labels: yarn-2928-1st-milestone > Attachments: YARN-4712-YARN-2928.v1.001.patch, > YARN-4712-YARN-2928.v1.002.patch, YARN-4712-YARN-2928.v1.003.patch, > YARN-4712-YARN-2928.v1.004.patch > > > There are 2 issues with CPU usage collection > * I was able to observe that that many times CPU usage got from > {{pTree.getCpuUsagePercent()}} is > ResourceCalculatorProcessTree.UNAVAILABLE(i.e. -1) but ContainersMonitor do > the calculation i.e. {{cpuUsageTotalCoresPercentage = cpuUsagePercentPerCore > /resourceCalculatorPlugin.getNumProcessors()}} because of which UNAVAILABLE > check in {{NMTimelinePublisher.reportContainerResourceUsage}} is not > encountered. so proper checks needs to be handled > * {{EntityColumnPrefix.METRIC}} uses always LongConverter but > ContainerMonitor is publishing decimal values for the CPU usage. -- This message was sent by Atlassian JIRA (v6.3.4#6332)