[ https://issues.apache.org/jira/browse/YARN-6862?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16098603#comment-16098603 ]
YunFan Zhou commented on YARN-6862: ----------------------------------- [~sunilg] Thanks. We only can see used memory from NM logs, and from NM logs we can see some logs as follows: 2017-07-24 22:19:08,551 INFO org.apache.hadoop.yarn.server.nodemanager.containermanager.monitor.ContainersMonitorImpl: Memory usage of ProcessTree 23933 for container-id container_e6717_1500903083707_0014_01_000259: -1B of 1 GB physical memory used; -1B of 2.1 GB virtual memory used Because we collect resource usage metrics direct from MonitoringThread#run method, so the metrics is very reliable. > Nodemanager resource usage metrics sometimes are negative > --------------------------------------------------------- > > Key: YARN-6862 > URL: https://issues.apache.org/jira/browse/YARN-6862 > Project: Hadoop YARN > Issue Type: Bug > Components: nodemanager > Affects Versions: 2.8.2 > Reporter: YunFan Zhou > > When we collect real-time metrics of resource usage in NM, we found those > values sometimes are invalid. > For example, the following are values when collected at some point: > "milliVcoresUsed":-5808, > "currentPmemUsage":-1, > "currentVmemUsage":-1, > "cpuUsagePercentPerCore":-968.1026 > "cpuUsageTotalCoresPercentage":-24.202564, > "pmemLimit":2147483648, > "vmemLimit":4509715456 > There are many negative values, there may a bug in NM. > We should fix it, because the real-time metrics of NM is pretty important for > us sometimes. -- This message was sent by Atlassian JIRA (v6.4.14#64029) --------------------------------------------------------------------- To unsubscribe, e-mail: yarn-issues-unsubscr...@hadoop.apache.org For additional commands, e-mail: yarn-issues-h...@hadoop.apache.org