[YARN-11053] AuxService should not use class name as default system classes

• Type: Bug
• Status: Resolved
• Priority: Major
• Resolution: Fixed
• Affects Version/s: 3.3.1
• Fix Version/s: 3.4.0, 3.3.2
• Component/s: auxservices
• Labels: pull-request-available
• Assignee: Cheng Pan
• Reporter: Cheng Pan
• Created: 18/Dec/21 12:35
• Updated: 24/Nov/22 00:38
• Resolved: 24/Dec/21 02:14

Description

Following Apache Spark document to configure Spark Shuffle Service as YARN AuxService,
https://spark.apache.org/docs/3.2.0/running-on-yarn.html#running-multiple-versions-of-the-spark-shuffle-service
 

  <property>
    <name>yarn.nodemanager.aux-services</name>
    <value>spark_shuffle</value>
  </property>
  <property>
    <name>yarn.nodemanager.aux-services.spark_shuffle.classpath</name>
    <value>/opt/apache/spark/yarn/*</value>
  </property>
  <property>
    <name>yarn.nodemanager.aux-services.spark_shuffle.class&lt;/name>
    <value>org.apache.spark.network.yarn.YarnShuffleService</value>
  </property> 

 but failed with exception

2021-12-02 15:34:00,886 INFO util.ApplicationClassLoader: classpath: [file:/opt/apache/spark/yarn/spark-3.2.0-yarn-shuffle.jar]
2021-12-02 15:34:00,886 INFO util.ApplicationClassLoader: system classes: [org.apache.spark.network.yarn.YarnShuffleService]
2021-12-02 15:34:00,887 INFO service.AbstractService: Service org.apache.hadoop.yarn.server.nodemanager.containermanager.AuxServices failed in state INITED
org.apache.hadoop.yarn.exceptions.YarnRuntimeException: java.lang.ClassNotFoundException: org.apache.spark.network.yarn.YarnShuffleService
        at org.apache.hadoop.yarn.server.nodemanager.containermanager.AuxServices.initAuxService(AuxServices.java:482)
        at org.apache.hadoop.yarn.server.nodemanager.containermanager.AuxServices.serviceInit(AuxServices.java:761)
        at org.apache.hadoop.service.AbstractService.init(AbstractService.java:164)
        at org.apache.hadoop.service.CompositeService.serviceInit(CompositeService.java:109)
        at org.apache.hadoop.yarn.server.nodemanager.containermanager.ContainerManagerImpl.serviceInit(ContainerManagerImpl.java:327)
        at org.apache.hadoop.service.AbstractService.init(AbstractService.java:164)
        at org.apache.hadoop.service.CompositeService.serviceInit(CompositeService.java:109)
        at org.apache.hadoop.yarn.server.nodemanager.NodeManager.serviceInit(NodeManager.java:494)
        at org.apache.hadoop.service.AbstractService.init(AbstractService.java:164)
        at org.apache.hadoop.yarn.server.nodemanager.NodeManager.initAndStartNodeManager(NodeManager.java:962)
        at org.apache.hadoop.yarn.server.nodemanager.NodeManager.main(NodeManager.java:1042)
Caused by: java.lang.ClassNotFoundException: org.apache.spark.network.yarn.YarnShuffleService
        at java.net.URLClassLoader.findClass(URLClassLoader.java:387)
        at java.lang.ClassLoader.loadClass(ClassLoader.java:419)
        at sun.misc.Launcher$AppClassLoader.loadClass(Launcher.java:352)
        at java.lang.ClassLoader.loadClass(ClassLoader.java:352)
        at org.apache.hadoop.util.ApplicationClassLoader.loadClass(ApplicationClassLoader.java:189)
        at org.apache.hadoop.util.ApplicationClassLoader.loadClass(ApplicationClassLoader.java:157)
        at java.lang.Class.forName0(Native Method)
        at java.lang.Class.forName(Class.java:348)
        at org.apache.hadoop.yarn.server.nodemanager.containermanager.AuxiliaryServiceWithCustomClassLoader.getInstance(AuxiliaryServiceWithCustomClassLoader.java:165)
        at org.apache.hadoop.yarn.server.nodemanager.containermanager.AuxServices.createAuxServiceFromLocalClasspath(AuxServices.java:242)
        at org.apache.hadoop.yarn.server.nodemanager.containermanager.AuxServices.createAuxService(AuxServices.java:271)
        at org.apache.hadoop.yarn.server.nodemanager.containermanager.AuxServices.initAuxService(AuxServices.java:452)
        ... 10 more

A workaround is adding

<property>
    <name>yarn.nodemanager.aux-services.spark_shuffle.system-classes</name>
    <value>not.existed.class&amp;amp;lt;/value>
 </property>

 

Issue Links

• duplicates: YARN-9967 Fix NodeManager failing to start when Hdfs Auxillary Jar is set
• is caused by: YARN-9075 Dynamically add or remove auxiliary services
• relates to: SPARK-37925 Update document to mention the workaround for YARN-11053
• links to: GitHub Pull Request #3816