当前位置: 首页>>代码示例>>Java>>正文


Java QueueMetrics.getAvailableGPUs方法代码示例

本文整理汇总了Java中org.apache.hadoop.yarn.server.resourcemanager.scheduler.QueueMetrics.getAvailableGPUs方法的典型用法代码示例。如果您正苦于以下问题:Java QueueMetrics.getAvailableGPUs方法的具体用法?Java QueueMetrics.getAvailableGPUs怎么用?Java QueueMetrics.getAvailableGPUs使用的例子?那么恭喜您, 这里精选的方法代码示例或许可以为您提供帮助。您也可以进一步了解该方法所在org.apache.hadoop.yarn.server.resourcemanager.scheduler.QueueMetrics的用法示例。


在下文中一共展示了QueueMetrics.getAvailableGPUs方法的3个代码示例,这些例子默认根据受欢迎程度排序。您可以为喜欢或者感觉有用的代码点赞,您的评价将有助于系统推荐出更棒的Java代码示例。

示例1: ClusterMetricsInfo

import org.apache.hadoop.yarn.server.resourcemanager.scheduler.QueueMetrics; //导入方法依赖的package包/类
public ClusterMetricsInfo(final ResourceManager rm) {
  ResourceScheduler rs = rm.getResourceScheduler();
  QueueMetrics metrics = rs.getRootQueueMetrics();
  ClusterMetrics clusterMetrics = ClusterMetrics.getMetrics();

  this.appsSubmitted = metrics.getAppsSubmitted();
  this.appsCompleted = metrics.getAppsCompleted();
  this.appsPending = metrics.getAppsPending();
  this.appsRunning = metrics.getAppsRunning();
  this.appsFailed = metrics.getAppsFailed();
  this.appsKilled = metrics.getAppsKilled();

  this.reservedMB = metrics.getReservedMB();
  this.availableMB = metrics.getAvailableMB();
  this.allocatedMB = metrics.getAllocatedMB();

  this.reservedVirtualCores = metrics.getReservedVirtualCores();
  this.availableVirtualCores = metrics.getAvailableVirtualCores();
  this.allocatedVirtualCores = metrics.getAllocatedVirtualCores();

  this.reservedGPUs = metrics.getReservedGPUs();
  this.availableGPUs = metrics.getAvailableGPUs();
  this.allocatedGPUs = metrics.getAllocatedGPUs();

  this.containersAllocated = metrics.getAllocatedContainers();
  this.containersPending = metrics.getPendingContainers();
  this.containersReserved = metrics.getReservedContainers();

  if (rs instanceof CapacityScheduler) {
    this.totalMB = availableMB + allocatedMB + reservedMB;
    this.totalVirtualCores = availableVirtualCores + allocatedVirtualCores
        + containersReserved;
    this.totalGPUs = availableGPUs + allocatedGPUs + reservedGPUs;
  } else {
    this.totalMB = availableMB + allocatedMB;
    this.totalVirtualCores = availableVirtualCores + allocatedVirtualCores;
    this.totalGPUs = availableGPUs + allocatedGPUs;
  }
  this.activeNodes = clusterMetrics.getNumActiveNMs();
  this.lostNodes = clusterMetrics.getNumLostNMs();
  this.unhealthyNodes = clusterMetrics.getUnhealthyNMs();
  this.decommissioningNodes = clusterMetrics.getNumDecommissioningNMs();
  this.decommissionedNodes = clusterMetrics.getNumDecommisionedNMs();
  this.rebootedNodes = clusterMetrics.getNumRebootedNMs();
  this.shutdownNodes = clusterMetrics.getNumShutdownNMs();
  this.totalNodes = activeNodes + lostNodes + decommissionedNodes
      + rebootedNodes + unhealthyNodes + shutdownNodes;
}
 
开发者ID:hopshadoop,项目名称:hops,代码行数:49,代码来源:ClusterMetricsInfo.java

示例2: getPercenUsedGpus

import org.apache.hadoop.yarn.server.resourcemanager.scheduler.QueueMetrics; //导入方法依赖的package包/类
private float getPercenUsedGpus(QueueMetrics metrics) {
  int totalGpus = metrics.getAllocatedGPUs() + metrics.getAvailableGPUs();
  int usedGpus = metrics.getAllocatedGPUs() + metrics.getPendingGPUs();
  return (float) usedGpus / totalGpus;
}
 
开发者ID:hopshadoop,项目名称:hops,代码行数:6,代码来源:PriceMultiplicatiorService.java

示例3: verifyClusterMetrics

import org.apache.hadoop.yarn.server.resourcemanager.scheduler.QueueMetrics; //导入方法依赖的package包/类
public void verifyClusterMetrics(int submittedApps, int completedApps,
    int reservedMB, int availableMB, int allocMB, int reservedVirtualCores,
    int availableVirtualCores, int allocVirtualCores, int totalVirtualCores,
    int reservedGPUs, int availableGPUs, int allocGPUs, int totalGPUs,
    int containersAlloc, int totalMB, int totalNodes, int lostNodes,
    int unhealthyNodes, int decommissionedNodes, int rebootedNodes,
    int activeNodes, int shutdownNodes) throws JSONException, Exception {

  ResourceScheduler rs = rm.getResourceScheduler();
  QueueMetrics metrics = rs.getRootQueueMetrics();
  ClusterMetrics clusterMetrics = ClusterMetrics.getMetrics();

  long totalMBExpect = 
      metrics.getAvailableMB() + metrics.getAllocatedMB();
  long totalVirtualCoresExpect = 
      metrics.getAvailableVirtualCores() + metrics.getAllocatedVirtualCores();
  long totalGPUsExpect =
      metrics.getAvailableGPUs() + metrics.getAllocatedGPUs();
  assertEquals("appsSubmitted doesn't match", 
      metrics.getAppsSubmitted(), submittedApps);
  assertEquals("appsCompleted doesn't match", 
      metrics.getAppsCompleted(), completedApps);
  assertEquals("reservedMB doesn't match",
      metrics.getReservedMB(), reservedMB);
  assertEquals("availableMB doesn't match", 
      metrics.getAvailableMB(), availableMB);
  assertEquals("allocatedMB doesn't match", 
      metrics.getAllocatedMB(), allocMB);
  assertEquals("reservedVirtualCores doesn't match",
      metrics.getReservedVirtualCores(), reservedVirtualCores);
  assertEquals("availableVirtualCores doesn't match",
      metrics.getAvailableVirtualCores(), availableVirtualCores);
  assertEquals("allocatedVirtualCores doesn't match",
      totalVirtualCoresExpect, allocVirtualCores);
  assertEquals("reservedGPUs doesn't match",
      metrics.getReservedGPUs(), reservedGPUs);
  assertEquals("availableGPUs doesn't match",
      metrics.getAvailableGPUs(), availableGPUs);
  assertEquals("allocatedGPUs doesn't match",
      totalGPUsExpect, allocGPUs);
  assertEquals("containersAllocated doesn't match", 0, containersAlloc);
  assertEquals("totalMB doesn't match", totalMBExpect, totalMB);
  assertEquals(
      "totalNodes doesn't match",
      clusterMetrics.getNumActiveNMs() + clusterMetrics.getNumLostNMs()
          + clusterMetrics.getNumDecommisionedNMs()
          + clusterMetrics.getNumRebootedNMs()
          + clusterMetrics.getUnhealthyNMs(), totalNodes);
  assertEquals("lostNodes doesn't match", clusterMetrics.getNumLostNMs(),
      lostNodes);
  assertEquals("unhealthyNodes doesn't match",
      clusterMetrics.getUnhealthyNMs(), unhealthyNodes);
  assertEquals("decommissionedNodes doesn't match",
      clusterMetrics.getNumDecommisionedNMs(), decommissionedNodes);
  assertEquals("rebootedNodes doesn't match",
      clusterMetrics.getNumRebootedNMs(), rebootedNodes);
  assertEquals("activeNodes doesn't match", clusterMetrics.getNumActiveNMs(),
      activeNodes);
  assertEquals("shutdownNodes doesn't match",
      clusterMetrics.getNumShutdownNMs(), shutdownNodes);
}
 
开发者ID:hopshadoop,项目名称:hops,代码行数:62,代码来源:TestRMWebServices.java


注:本文中的org.apache.hadoop.yarn.server.resourcemanager.scheduler.QueueMetrics.getAvailableGPUs方法示例由纯净天空整理自Github/MSDocs等开源代码及文档管理平台,相关代码片段筛选自各路编程大神贡献的开源项目,源码版权归原作者所有,传播和使用请参考对应项目的License;未经允许,请勿转载。