Hadoop Yarn Capacity Scheduler

 

 

Capacity 調度器配置

<property>
  <name>yarn.resourcemanager.scheduler.class<name>
  <value>org.apache.hadoop.yarn.server.resourcemanager.scheduler.capacity.CapacityScheduler</value>
</property>

 

 

加載配置文件

$yarn rmadmin -refreshQueues

 

 

Root --- QueueA --- A1,  A2
          |-- QueueB --- B1,  B2
          |-- QueueC --- C1,  C2

 

 

隊列

<property>
  <name>yarn.scheduler.capacity.root.queues<name>
  <value>QueueA, QueueB, QueueC</value>
</property>

<property>
  <name>yarn.scheduler.capacity.QueueA.queues<name>
  <value>A1, A2</value>
</property>

 

 

隊列訪問權限控制

<property>
  <name>yarn.scheduler.capacity.root.QueueA.acl_submit_applications<name>
  <value>u1, u2 g1</value>
  <description>限定哪些Linux用戶/用戶組可向給定隊列中提交應用程序</description>
</property>

<property>
  <name>yarn.scheduler.capacity.root.QueueA.acl_administer_queue<name>
  <value>g1</value>
  <description>爲隊列指定一個管理員，該管理員可控制該隊列的全部應用程序，好比殺死任意一個應用程序等</description>
</property>

 

 

層級隊列Capacity管理

<property>
  <name>yarn.scheduler.capacity.root.QueueA.capacity<name>
  <value>60</value>
</property>

<property>
  <name>yarn.scheduler.capacity.root.QueueB.capacity<name>
  <value>10</value>
</property>

<property>
  <name>yarn.scheduler.capacity.root.QueueC.capacity<name>
  <value>30</value>
</property>

<property>
  <name>yarn.scheduler.capacity.root.QueueA.A1.capacity<name>
  <value>20</value>
</property>

<property>
  <name>yarn.scheduler.capacity.root.QueueA.A2.capacity<name>
  <value>80</value>
</property>

 

 

隊列A1不能超過父隊列queueA的40%的資源。

<property>
  <name>yarn.scheduler.capacity.root.QueueA.A1.maximum-capacity<name>
  <value>40</value>
</property>

 

 

隊列的狀態

<property>
  <name>yarn.scheduler.capacity.root.QueueB.state<name>
  <value>RUNNING</value>
</property>

 

 

應用程序限制

<property>
  <name>yarn.scheduler.capacity.maximum-applications<name>
  <value>10000</value>
  <description>集羣同時運行的最大做業數</description>
</property>

<property>
  <name>yarn.scheduler.capacity.root.QueueA.maximum-applications<name>
  <value>10000</value>
  <description>隊列root.QueueA同時運行的最大做業數</description>
</property>

 

<property>
  <name>yarn.scheduler.capacity.maximum-am-resource-percent<name>
  <value>0.1</value>
  <description>AM所佔資源最大比例，用來控制同時運行的做業數</description>
</property>

<property>
  <name>yarn.scheduler.capacity.root.QueueA.maximum-am-resource-percent<name>
  <value>0.1</value>
  <description>AM所佔資源最大比例，用來控制隊列root.QueueA同時運行的做業數</description>
</property>