Oozie部署spark、hadoop、hive、shell和java程序举例

超、凢脫俗 2022-05-15 03:44 153阅读 0赞

一、oozie执行shell脚本(执行mr任务,实现合并增量数据)

**参考**：[http://gethue.com/use-the-shell-action-in-oozie/][http_gethue.com_use-the-shell-action-in-oozie]

1、点击创建、拖动到上面

![Center][]

2、添加命令:bash，当然也可以是linux的其他可执行的命令

![Center 1][]

3、添加参数：注意---》参数名称是shell脚本的全称(run-mr-compact.sh)

![Center 2][]

4、添加xxx.sh、xxx.jar、xxx.properties文件（注意：这些文件需要在同一个hdfs文件夹目录下，否则会报错！）

![Center 3][]

添加的文件顺序建议是：sh 、jar 、properties

![Center 4][]

相关文件内容：

(1)/user/greatgas/oozie/everyday/shell/run-mr-compact.sh.sh：

<table> 
 <tbody> 
  <tr> 
   <td style="vertical-align:top;"> <p>&nbsp;</p> <p>#!/bin/bash</p> <p>hadoop jar mr-compact.jar main类全类名&nbsp;&nbsp; properties文件相对于该jar的路径</p> <p>&nbsp;</p> </td> 
  </tr> 
 </tbody> 
</table>

(2)/user/greatgas/oozie/everyday/MR/mr-compact.jar：

<table> 
 <tbody> 
  <tr> 
   <td style="vertical-align:top;"> <p>MR合并程序</p> </td> 
  </tr> 
 </tbody> 
</table>

(3)/user/greatgas/oozie/everyday/MR/conf.properties：

<table> 
 <tbody> 
  <tr> 
   <td style="vertical-align:top;"> <p>jobName=t_test</p> <p>&nbsp;</p> <p>#baseDir=hdfs上数据库路径</p> <p>baseDir=/user/hive/warehouse/origin_ennenergy_test.db/</p> <p>&nbsp;</p> <p>#就是根据table name来寻找hdfs中的hive数据路径</p> <p>#tableName=原始表,增量表,中转表</p> <p>tableName= s_t_test,incr_ t_test,out_ t_test</p> <p>&nbsp;</p> <p>keyIndex=0</p> <p>#时间戳所在字段，用于判断哪条记录是最新的</p> <p>timeStampIndex=24</p> <p>reduceNum=2</p> <p>&nbsp;</p> </td> 
  </tr> 
 </tbody> 
</table>

5、点击红框添加属性

![Center 5][]

6、属性添加

![Center 6][]

![Center 7][]

注意：一定要添加”HADOOP\_USER\_NAME=hue用户名”，否则没有执行hdfs的文件目录权限！

7、点击6图片中的右上角，退出，然后保存

![Center 8][]

8、提交任务

![Center 9][]

二、oozie执行mapreduce任务(实现合并增量数据)

需要上传2个文件到hdfs(如果是集群模式执行)

![Center 10][]

/user/e\_test/workflow/mr/mr-compact.jar(hdfslinux目录jar)

mapred.output.dir   /user/hive/warehouse/origin\_ennenergy\_onecard.db/m(输出路径)

mapred.input.dir     /user/hive/warehouse/origin\_ennenergy\_onecard.db/t(输入路径)

delete                       /user/hive/warehouse/origin\_ennenergy\_onecard.db/m(删除)

/user/e\_test/workflow/mr/conf.properties(hdfslinux目录properties)

![Center 11][]

编辑完成后点击提交即可！

三、oozie执行spark任务(实现实时接收kafka消息队列的数据)

1、创建oozie项目

![Center 12][]

2、添加参数：

yarn-cluster  (相当于setMaster(“yarn-cluster”))

cluster     (client或者cluster模式，一般都是使用cluster)

MySpark   (job名称，可以任意写)

hdfs://master-28.dev.cluster.enn.cn:8020/user/e\_liuy/testspark/spark-test.jar(hdfs路径)

或者

hdfs:// nameservice/user/e\_liuy/testspark/spark-test.jar(注意集群总名称可以替代url+端口)

enn.action.ConsumerRealDataKafka   (需要运行的类的全路径)

四、oozie执行hive任务(实现在hive的origin\_ennenergy\_onecard数据库中创建一个表----->M\_BD\_CD\_CarInfo\_H)

1、创建oozie的hive项目

2、添加sh脚本和hive xml文件

![Center 13][]

文件内容

(1)[onecardtomodel.q][]/

use origin\_ennenergy\_onecard;

drop table if exists M\_BD\_CD\_CarInfo\_H;

create table M\_BD\_CD\_CarInfo\_H

(

FGUID                STRING,

FStationNo           STRING,

FCOMPANYID           STRING,

FCARNO               STRING,

FCARER               STRING,

FTEL                 STRING,

time\_stamp           STRING,

)

ROW FORMAT DELIMITED FIELDS TERMINATED BY'\\t'

STORED AS TEXTFILE;

(2)hive-site.xml

内容：

1.  `<?xml version="1.0" encoding="UTF-8"?>`
2.   
3.   
4.  `hive.metastore.uristhrift://yours_hosts:9083hive.metastore.client.socket.timeout300hive.metastore.warehouse.dir/user/hive/warehouse(hive的hdfs目录)hive.warehouse.subdir.inherit.permstruehive.enable.spark.execution.enginefalsehive.conf.restricted.listhive.enable.spark.execution.enginehive.auto.convert.jointruehive.auto.convert.join.noconditionaltask.size20971520hive.optimize.bucketmapjoin.sortedmergefalsehive.smbjoin.cache.rows10000mapred.reduce.tasks-1hive.exec.reducers.bytes.per.reducer67108864hive.exec.copyfile.maxsize33554432hive.vectorized.groupby.checkinterval4096hive.vectorized.groupby.flush.percent0.1hive.compute.query.using.statsfalsehive.vectorized.execution.enabledtruehive.vectorized.execution.reduce.enabledfalsehive.merge.mapfilestruehive.merge.mapredfilesfalsehive.cbo.enablefalsehive.fetch.task.conversionminimalhive.fetch.task.conversion.threshold268435456hive.limit.pushdown.memory.usage0.1hive.merge.sparkfilestruehive.merge.smallfiles.avgsize16777216hive.merge.size.per.task268435456hive.optimize.reducededuplicationtruehive.optimize.reducededuplication.min.reducer4hive.map.aggrtruehive.map.aggr.hash.percentmemory0.5hive.optimize.sort.dynamic.partitionfalsespark.executor.memory268435456spark.driver.memory268435456spark.executor.cores1spark.yarn.driver.memoryOverhead26spark.yarn.executor.memoryOverhead26spark.dynamicAllocation.enabledtruespark.dynamicAllocation.initialExecutors1spark.dynamicAllocation.minExecutors1spark.dynamicAllocation.maxExecutors2147483647hive.metastore.execute.setugitruehive.support.concurrencytruehive.zookeeper.quorumslave-29.dev.cluster.enn.cn,slave-30.dev.cluster.enn.cn,slave-31.dev.cluster.enn.cnhive.zookeeper.client.port2181hive.zookeeper.namespacehive_zookeeper_namespace_hive2hbase.zookeeper.quorumzookeeper的机器列表（node1,node2,node3……）hbase.zookeeper.property.clientPort2181hive.cluster.delegation.token.store.classorg.apache.hadoop.hive.thrift.MemoryTokenStorehive.server2.enable.doAstruehive.metastore.sasl.enabledtruehive.server2.authenticationkerberoshive.metastore.kerberos.principalhive/_HOST@ENN.CNhive.server2.authentication.kerberos.principalhive/_HOST@ENN.CNspark.shuffle.service.enabledtruehive.cli.print.current.dbtruehive.exec.reducers.max32`

3、保存项目后，提交任务即可

五、oozie执行sqoop任务（实现将hive数据导出到mysql中）

1、创建oozie的sqoop项目

![Center 14][]

2、输入sqoop命令

![Center 15][]

命令内容：

export --connectjdbc:mysql://10.37.149.183:3306/enn\_application?characterEncoding=utf8--username root --password secret\_password --table total\_info --export-dir/user/hive/warehouse/origin\_ennenergy\_onecard.db/total\_info/\*--input-fields-terminated-by "\\t" --update-mode allowinsert--update-key fstationno,fusercardno

3、保存执行即可

**六、添加java程序**

1、上传jar到liunx对应的oozie目录路径上

2、开始在Hue上操作：

![Center 16][]

**以上测试已经通过，可以按要求更改后直接使用！**

[http_gethue.com_use-the-shell-action-in-oozie]: http://gethue.com/use-the-shell-action-in-oozie/
[Center]: /images/20220515/dbb12c6a20684fe794cde6ddcb7553fd.png
[Center 1]: /images/20220515/1d53c913a52b474f9720b9ee2de39394.png
[Center 2]: /images/20220515/3b76f4ce90e64f989e7fbcf05e96522c.png
[Center 3]: /images/20220515/61866c5b411d45398e54deea9ee7f977.png
[Center 4]: /images/20220515/454acc74d99d427c880e3e16e516000f.png
[Center 5]: /images/20220515/6564cd1f339b43b6b5446f818235d094.png
[Center 6]: /images/20220515/fceadea2dd094eb0abd52b01f71d52f4.png
[Center 7]: /images/20220515/6e65fabe0a3c49d9b57b2f6a9b49a109.png
[Center 8]: /images/20220515/3641c7c3c1ce45c399a72c3723a9df6d.png
[Center 9]: /images/20220515/b3c2d860e43641c58b821939809045b5.png
[Center 10]: /images/20220515/806fd2934e94444796e024d056e38671.png
[Center 11]: /images/20220515/5c276ddbfbb74bdb90f12e383956d0f0.png
[Center 12]: /images/20220515/4ac31627aa3a4a25940a8355ac8c10f8.png
[Center 13]: /images/20220515/cbcb44f94e5c463292b0a0f431858210.png
[onecardtomodel.q]: http://master-27.dev.cluster.enn.cn:8888/filebrowser/view=/user/e_heyutao/workflow/hive/onecardtomodel.q
[Center 14]: /images/20220515/ae97192feeb4423190c44df64633e48b.png
[Center 15]: /images/20220515/e9e4926e11aa426eb99252afbd5e3f34.png
[Center 16]: /images/20220515/d19a4d85381f4deb8a7e0daa0cf59b6a.png