[KYLIN-4005] Saving Cube of a aggregation Groups(40 Dimensions, Max Dimension Combination:5) may cause kylin server OOM

• Type: Bug
• Status: Closed
• Priority: Critical
• Resolution: Fixed
• Affects Version/s: v2.5.2
• Fix Version/s: v3.0.0-alpha2
• Component/s: REST Service
• Assignee: Shaohui Liu
• Reporter: Shaohui Liu
• Created: 16/May/19 10:50
• Updated: 21/Jan/20 07:43
• Resolved: 17/Jul/19 07:10

Description

A user try to save a cube with a aggregation Groups(40 Dimensions, Max Dimension Combination:5) caused the kylin server OOM. The reason is that the DefaultCuboidScheduler will cost a lot memory when calculating all cube ids. The stack is following

http-bio-7070-exec-35
  at java.lang.OutOfMemoryError.<init>()V (OutOfMemoryError.java:48)
  at java.util.HashMap.resize()[Ljava/util/HashMap$Node; (HashMap.java:704)
  at java.util.HashMap.putVal(ILjava/lang/Object;Ljava/lang/Object;ZZ)Ljava/lang/Object; (HashMap.java:663)
  at java.util.HashMap.put(Ljava/lang/Object;Ljava/lang/Object;)Ljava/lang/Object; (HashMap.java:612)
  at java.util.HashSet.add(Ljava/lang/Object;)Z (HashSet.java:220)
  at java.util.AbstractCollection.addAll(Ljava/util/Collection;)Z (AbstractCollection.java:344)
  at org.apache.kylin.cube.cuboid.DefaultCuboidScheduler.getOnTreeParentsByLayer(Ljava/util/Collection;)Ljava/util/Set; (DefaultCuboidScheduler.java:240)
  at org.apache.kylin.cube.cuboid.DefaultCuboidScheduler.buildTreeBottomUp()Lorg/apache/kylin/common/util/Pair; (DefaultCuboidScheduler.java:183)
  at org.apache.kylin.cube.cuboid.DefaultCuboidScheduler.<init>(Lorg/apache/kylin/cube/model/CubeDesc;)V (DefaultCuboidScheduler.java:58)
  at sun.reflect.GeneratedConstructorAccessor140.newInstance([Ljava/lang/Object;)Ljava/lang/Object; (Unknown Source)
  at sun.reflect.DelegatingConstructorAccessorImpl.newInstance([Ljava/lang/Object;)Ljava/lang/Object; (DelegatingConstructorAccessorImpl.java:45)
  at java.lang.reflect.Constructor.newInstance([Ljava/lang/Object;)Ljava/lang/Object; (Constructor.java:423)
  at org.apache.kylin.cube.cuboid.CuboidScheduler.getInstance(Lorg/apache/kylin/cube/model/CubeDesc;)Lorg/apache/kylin/cube/cuboid/CuboidScheduler; (CuboidScheduler.java:41)
  at org.apache.kylin.cube.model.CubeDesc.getInitialCuboidScheduler()Lorg/apache/kylin/cube/cuboid/CuboidScheduler; (CubeDesc.java:750)
  at org.apache.kylin.cube.cuboid.CuboidCLI.simulateCuboidGeneration(Lorg/apache/kylin/cube/model/CubeDesc;Z)I (CuboidCLI.java:47)
  at org.apache.kylin.rest.service.CubeService.updateCubeAndDesc(Lorg/apache/kylin/cube/CubeInstance;Lorg/apache/kylin/cube/model/CubeDesc;Ljava/lang/String;Z)Lorg/apache/kylin/cube/model/CubeDesc; (CubeService.java:287)
  at org.apache.kylin.rest.service.CubeService$$FastClassBySpringCGLIB$$17a07c0e.invoke(ILjava/lang/Object;[Ljava/lang/Object;)Ljava/lang/Object; (Unknown Source)
  at org.springframework.cglib.proxy.MethodProxy.invoke(Ljava/lang/Object;[Ljava/lang/Object;)Ljava/lang/Object; (MethodProxy.java:204)
  at org.springframework.aop.framework.CglibAopProxy$DynamicAdvisedInterceptor.intercept(Ljava/lang/Object;Ljava/lang/reflect/Method;[Ljava/lang/Object;Lorg/springframework/cglib/proxy/MethodProxy;)Ljava/lang/Object; (CglibAopProxy.java:669)
  at org.apache.kylin.rest.service.CubeService$$EnhancerBySpringCGLIB$$34de75c4.updateCubeAndDesc(Lorg/apache/kylin/cube/CubeInstance;Lorg/apache/kylin/cube/model/CubeDesc;Ljava/lang/String;Z)Lorg/apache/kylin/cube/model/CubeDesc; (Unknown Source)
  at org.apache.kylin.rest.controller.CubeController.updateCubeDesc(Lorg/apache/kylin/rest/request/CubeRequest;)Lorg/apache/kylin/rest/request/CubeReq

Issue Links

• links to: GitHub Pull Request #668