执行join操作时localtask启动失败

问题背景与现象

1. 执行join等操作，数据量较小时，会启动localtask执行，执行过程会报错：

   jdbc:hive2://10.*.*.*:21066/> select a.name ,b.sex from student a join student1 b on (a.name = b.name);
   ERROR : Execution failed with exit status: 1
   ERROR : Obtaining error information
   ERROR : 
   Task failed!
   Task ID:
     Stage-4
   ...
   Error: Error while processing statement: FAILED: Execution Error, return code 1 from org.apache.hadoop.hive.ql.exec.mr.MapredLocalTask (state=08S01,code=1)
   ...

2. 查看对应HiveServer日志，发现是启动localtask失败：

   2018-04-25 16:37:19,296 | ERROR | HiveServer2-Background-Pool: Thread-79 | Execution failed with exit status: 1 | org.apache.hadoop.hive.ql.session.SessionState$LogHelper.printError(SessionState.java:1016)
   2018-04-25 16:37:19,296 | ERROR | HiveServer2-Background-Pool: Thread-79 | Obtaining error information | org.apache.hadoop.hive.ql.session.SessionState$LogHelper.printError(SessionState.java:1016)
   2018-04-25 16:37:19,297 | ERROR | HiveServer2-Background-Pool: Thread-79 |
   Task failed!
   Task ID:
     Stage-4
   Logs:
    | org.apache.hadoop.hive.ql.session.SessionState$LogHelper.printError(SessionState.java:1016)
   2018-04-25 16:37:19,297 | ERROR | HiveServer2-Background-Pool: Thread-79 | /var/log/Bigdata/hive/hiveserver/hive.log | org.apache.hadoop.hive.ql.session.SessionState$LogHelper.printError(SessionState.java:1016)
   2018-04-25 16:37:19,297 | ERROR | HiveServer2-Background-Pool: Thread-79 | Execution failed with exit status: 1 | org.apache.hadoop.hive.ql.exec.mr.MapredLocalTask.executeInChildVM(MapredLocalTask.java:342)
   2018-04-25 16:37:19,309 | ERROR | HiveServer2-Background-Pool: Thread-79 | FAILED: Execution Error, return code 1 from org.apache.hadoop.hive.ql.exec.mr.MapredLocalTask | org.apache.hadoop.hive.ql.session.SessionState$LogHelper.printError(SessionState.java:1016)
   ...
   2018-04-25 16:37:36,438 | ERROR | HiveServer2-Background-Pool: Thread-88 | Error running hive query:  | org.apache.hive.service.cli.operation.SQLOperation$1$1.run(SQLOperation.java:248)
   org.apache.hive.service.cli.HiveSQLException: Error while processing statement: FAILED: Execution Error, return code 1 from org.apache.hadoop.hive.ql.exec.mr.MapredLocalTask
           at org.apache.hive.service.cli.operation.Operation.toSQLException(Operation.java:339)
           at org.apache.hive.service.cli.operation.SQLOperation.runQuery(SQLOperation.java:169)
           at org.apache.hive.service.cli.operation.SQLOperation.access$200(SQLOperation.java:75)
           at org.apache.hive.service.cli.operation.SQLOperation$1$1.run(SQLOperation.java:245)
           at java.security.AccessController.doPrivileged(Native Method)
           at javax.security.auth.Subject.doAs(Subject.java:422)
           at org.apache.hadoop.security.UserGroupInformation.doAs(UserGroupInformation.java:1710)
           at org.apache.hive.service.cli.operation.SQLOperation$1.run(SQLOperation.java:258)
           at java.util.concurrent.Executors$RunnableAdapter.call(Executors.java:511)
           at java.util.concurrent.FutureTask.run(FutureTask.java:266)
           at java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1142)
           at java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:617)
           at java.lang.Thread.run(Thread.java:745)

3. 查看对应HiveServer日志目录（/var/log/Bigdata/hive/hiveserver）下的hs_err_pid_*****.log，发现有内存不够的错误：

   # There is insufficient memory for the Java Runtime Environment to continue.
   # Native memory allocation (mmap) failed to map 20776943616 bytes for committing reserved memory.
     ...

原因分析

Hive在执行join操作，数据量小时会生成MapJoin，执行MapJoin时会生成localtask任务，localtask启动的jvm内存继承了父进程的内存。

当有多个join执行的时候，启动多个localtask，如果机器内存不够，就会导致启动localtask失败。

解决办法

1. 进入Hive服务配置页面：
   • MRS 3.x之前版本，单击集群名称，登录集群详情页面，选择“组件管理 > Hive > 服务配置”，单击“基础配置”下拉菜单，选择“全部配置”。
     

     若集群详情页面没有“组件管理”页签，请先完成IAM用户同步（在集群详情页的“概览”页签，单击“IAM用户同步”右侧的“同步”进行IAM用户同步）。

   • MRS 3.x及后续版本，登录FusionInsight Manager，然后选择“集群 > 待操作的集群名称 > 服务 > Hive > 配置 > 全部配置”。

2. 搜索“hive.auto.convert.join”参数并修改hive的配置hive.auto.convert.join为false，保存配置并重启服务。

   该参数修改后会对业务性能有一定影响。继续执行后续步骤可不影响业务性能。

3. 搜索“HIVE_GC_OPTS”参数并修改，把Xms调小，具体要根据业务评估，最小设置为Xmx的一半，修改完后保存配置并重启服务。