Question

我是Hive的新手，我正在使用HBASE-1.1.0，Hadoop-2.5.1和Hive-0.13来满足我的要求。

设置绝对正常，我可以使用beeline运行配置单元查询。

查询：从X_Table中选择计数（*）。

查询以37.848秒完成。

我使用Maven项目设置的相同环境，并尝试使用Hive Client执行一些选择查询。但是当我尝试执行相同的计数查询时，Mapreduce作业无法完成。它看起来像重新开始工作。我该如何解决这个问题？

代码

Connection con = DriverManager.getConnection("jdbc:hive2://abc:10000/default","", "");
    Statement stmt = con.createStatement();               
    String query = "select count(*) from X_Table                                    
    ResultSet res = stmt.executeQuery(query);       
    while (res.next()) {
        //code here
    }

日志详情：

Total jobs = 1 
Launching Job 1 out of 1
Number of reduce tasks determined at compile time: 1
In order to change the average load for a reducer (in bytes):  
    set hive.exec.reducers.bytes.per.reducer=<number>
In order to limit the maximum number of reducers:  
    set hive.exec.reducers.max=<number>
In order to set a constant number of reducers:
    set mapreduce.job.reduces=<number>
Starting Job = job_1429243611915_0030, Tracking URL = http://master:8088/proxy/application_1429243611915_0030/
Kill Command = /usr/local/pcs/hadoop/bin/hadoop job  -kill job_1429243611915_0030
Hadoop job information for Stage-1: number of mappers: 1; number of reducers: 1
2015-04-20 09:28:02,616 Stage-1 map = 0%,  reduce = 0%
2015-04-20 09:29:02,728 Stage-1 map = 0%,  reduce = 0%
2015-04-20 09:30:03,432 Stage-1 map = 0%,  reduce = 0%
2015-04-20 09:31:04,054 Stage-1 map = 0%,  reduce = 0%
2015-04-20 09:32:04,675 Stage-1 map = 0%,  reduce = 0%
2015-04-20 09:33:05,298 Stage-1 map = 0%,  reduce = 0%
2015-04-20 09:34:05,866 Stage-1 map = 0%,  reduce = 0%
2015-04-20 09:35:06,419 Stage-1 map = 0%,  reduce = 0%
2015-04-20 09:36:06,985 Stage-1 map = 0%,  reduce = 0%
2015-04-20 09:37:07,551 Stage-1 map = 0%,  reduce = 0%
2015-04-20 09:38:08,289 Stage-1 map = 0%,  reduce = 0%
2015-04-20 09:39:09,184 Stage-1 map = 0%,  reduce = 0%
2015-04-20 09:40:09,780 Stage-1 map = 0%,  reduce = 0%
2015-04-20 09:41:10,367 Stage-1 map = 0%,  reduce = 0%
2015-04-20 09:42:10,965 Stage-1 map = 0%,  reduce = 0%
2015-04-20 09:43:11,595 Stage-1 map = 0%,  reduce = 0%
2015-04-20 09:44:12,181 Stage-1 map = 0%,  reduce = 0%
2015-04-20 09:45:12,952 Stage-1 map = 0%,  reduce = 0%
2015-04-20 09:46:13,590 Stage-1 map = 0%,  reduce = 0%
2015-04-20 09:47:14,218 Stage-1 map = 0%,  reduce = 0%
2015-04-20 09:48:14,790 Stage-1 map = 0%,  reduce = 0%
2015-04-20 09:49:15,378 Stage-1 map = 0%,  reduce = 0%
2015-04-20 09:50:16,014 Stage-1 map = 0%,  reduce = 0%
2015-04-20 09:51:16,808 Stage-1 map = 0%,  reduce = 0%
2015-04-20 09:52:17,378 Stage-1 map = 0%,  reduce = 0%
2015-04-20 09:53:17,928 Stage-1 map = 0%,  reduce = 0%
2015-04-20 09:54:18,491 Stage-1 map = 0%,  reduce = 0%
2015-04-20 09:55:19,049 Stage-1 map = 0%,  reduce = 0%
2015-04-20 09:56:19,797 Stage-1 map = 0%,  reduce = 0%
2015-04-20 09:57:20,344 Stage-1 map = 0%,  reduce = 0%
Total jobs = 1
Launching Job 1 out of 1
Number of reduce tasks determined at compile time: 1
In order to change the average load for a reducer (in bytes):
    set hive.exec.reducers.bytes.per.reducer=<number>
In order to limit the maximum number of reducers:
    set hive.exec.reducers.max=<number>
In order to set a constant number of reducers:
    set mapreduce.job.reduces=<number>
Starting Job = job_1429243611915_0031, Tracking URL = http://master:8088/proxy/application_1429243611915_0031/
Kill Command = /usr/local/pcs/hadoop/bin/hadoop job  -kill job_1429243611915_0031
2015-04-20 09:58:20,858 Stage-1 map = 0%,  reduce = 0%

Answer 1

如果在yarn-site.xml文件中为这两个配置增加内存，那么它将快速运行。

yarn.scheduler.maximum-allocation-mb yarn.nodemanager.resource.memory-mb

Answer 2

上面的答案很有用，它对我帮助很大。我试图在HIVE中运行一个简单的count（*）查询，但它既不会出错也不会完成。它将挂在那里直到我在命令提示符中终止该作业。我完全疯了，我没有得到谷歌的适当答案。但上面的答案对我帮助很大。所以我们需要增加

的记忆

^help$
yarn.scheduler.maximum-allocation-mb

这可以在yarn.nodemanager.resource.memory-mb中完成，甚至可以在Yarn Service下的Cloudera Manager中完成。一旦增加内存，就重新启动所有过时的服务。这将解决问题。

hive count查询无法完成它永远运行

2 个答案: