Fork me on GitHub

Oozie

Oozie的功能模块

workflow
由多个工作单元组成
工作单元之间有依赖关系
MR1->MR2->MR3->result
hadoop jar:提交1个MR
oozie:监控当前工作单元状态,完成之后自动提交下一个工作单元
scheduler
crontab:是linux简单调度脚本
定时调度工作单元
模块:
1) workflow:定义工作流程;顺序执行流程节点,支持fork(分支多个节点),join(合并多个节点为一个)

2) Coordinator:协调器,调度工作流程,只能定义一个周期;定时触发workflow

3) Bundle:绑定多个定时任务;绑定多个Coordinator 

workflow常用节点:

1) 工作单元-对应->动作节点Action Nodes
2) 指明工作流走向->控制节点Control Flow Nodes

部署Hadoop(CDH版本的)

apache版本的oozie需要编译(即用maven导入相应的jar包)
Namenode:core-site.xml
hdfs:9000->8020为了区别之前的hadoop

core-site.xml

    <!-- 指定HDFS中NameNode的地址 -->
    <property>
        <name>fs.defaultFS</name>
        <value>hdfs://hadoop101:8020</value>
    </property>
    <!-- 指定Hadoop运行时产生文件的存储目录 -->
    <property>
        <name>hadoop.tmp.dir</name>
        <value>/opt/module/cdh/hadoop-2.5.0-cdh5.3.6</value>
    </property>
    <!-- Oozie Server的Hostname -->
    <property>
        <name>hadoop.proxyuser.kris.hosts</name>
        <value>*</value>
    </property>

    <!-- 允许被Oozie代理的用户组 -->
    <property>
        <name>hadoop.proxyuser.kris.groups</name>
        <value>*</value>
    </property>
View Code

hadoop代理用户
有可能没有权限,让代理去做无关紧要的事
kris被代理,user1代理;oozie用到了hadoop的代理机制

kris能够代理用户的节点
kris能够代理的用户组
即kris能够代理所有节点上的所有用户

hdfs-site.xml

    <!-- 数据的副本数量 -->
    <property>
        <name>dfs.replication</name>
        <value>3</value>
    </property>
    <!-- 指定Hadoop辅助名称节点主机配置 -->
    <property>
          <name>dfs.namenode.secondary.http-address</name>
          <value>hadoop103:50090</value>
    </property>
View Code

yarn-site.xml

    <!-- Reducer获取数据的方式 -->
    <property>
            <name>yarn.nodemanager.aux-services</name>
            <value>mapreduce_shuffle</value>
    </property>

    <!-- 指定YARN的ResourceManager的地址 -->
    <property>
        <name>yarn.resourcemanager.hostname</name>
        <value>hadoop102</value>
    </property>
    <!-- 任务历史服务 -->
    <property> 
        <name>yarn.log.server.url</name> 
        <value>http://hadoop101:19888/jobhistory/logs/</value> 
    </property>
View Code

改名为mapred-site.xml

    <property>
        <name>mapreduce.framework.name</name>
        <value>yarn</value>
    </property>
    <!-- 配置 MapReduce JobHistory Server 地址 ,默认端口10020 -->
    <property>
        <name>mapreduce.jobhistory.address</name>
        <value>hadoop101:10020</value>
    </property>

    <!-- 配置 MapReduce JobHistory Server web ui 地址, 默认端口19888 -->
    <property>
        <name>mapreduce.jobhistory.webapp.address</name>
        <value>hadoop101:19888</value>
    </property>    
View Code

oozie是通过历史服务器进行监控工作单元
服务地址端口10020,web页面的端口是19888

slaves

hadoop101
hadoop102
hadoop103
View Code

启动hadoop集群:
[kris@hadoop101 hadoop-2.5.0-cdh5.3.6]$bin/hdfs namenode -format ##如果出现reformat就点击N
[kris@hadoop101 hadoop-2.5.0-cdh5.3.6]$sbin/start-dfs.sh
[kris@hadoop102 hadoop-2.5.0-cdh5.3.6]$sbin/start-yarn.sh
[kris@hadoop101 hadoop-2.5.0-cdh5.3.6]$sbin/mr-jobhistory-daemon.sh start historyserver

部署Oozie

1. 解压Oozie
  [kris@hadoop101 software]$ tar -zxvf /opt/software/oozie-4.0.0-cdh5.3.6.tar.gz -C /opt/module
2. 在oozie根目录下解压oozie-hadooplibs-4.0.0-cdh5.3.6.tar.gz;因为它上级目录与这个目录一样,它就会直接把hadooplibs解压到根目录;
  [kris@hadoop101 oozie-4.0.0-cdh5.3.6]$ tar -zxvf oozie-hadooplibs-4.0.0-cdh5.3.6.tar.gz -C ../
    也可以: tar -zxf oozie-hadooplibs-4.0.0-cdh5.3.6.tar.gz 
           mv hadooplibs/ ../
  完成后Oozie目录下会出现hadooplibs目录。 3. 在Oozie目录下创建libext目录   [kris@hadoop101 oozie-4.0.0-cdh5.3.6]$ mkdir libext/ 4. 拷贝依赖的Jar包;jar包、mysql驱动类、js框架放到libext目录下;   1)将hadooplibs里面的jar包,拷贝到libext目录下:   [kris@hadoop101 oozie-4.0.0-cdh5.3.6]$ cp -ra hadooplibs/hadooplib-2.5.0-cdh5.3.6.oozie-4.0.0-cdh5.3.6/* libext/   2)拷贝Mysql驱动包到libext目录下:   [kris@hadoop101 oozie-4.0.0-cdh5.3.6]$ cp -a /opt/software/mysql-connector-java-5.1.27/mysql-connector-java-5.1.27-bin.jar ./libext/   3)将ext-2.2.zip拷贝到libext/目录下   ext是一个js框架,用于展示oozie前端页面:   [kris@hadoop101 oozie-4.0.0-cdh5.3.6]$ cp -a /opt/software/ext-2.2.zip libext/
5. 修改Oozie配置文件
oozie-site.xml
属性:oozie.service.JPAService.jdbc.driver
属性值:com.mysql.jdbc.Driver
解释:JDBC的驱动

属性:oozie.service.JPAService.jdbc.url
属性值:jdbc:mysql://hadoop101:3306/oozie
解释:oozie所需的数据库地址

属性:oozie.service.JPAService.jdbc.username
属性值:root
解释:数据库用户名

属性:oozie.service.JPAService.jdbc.password
属性值:123456
解释:数据库密码

属性:oozie.service.HadoopAccessorService.hadoop.configurations
属性值:*=/opt/module/cdh/hadoop-2.5.0-cdh5.3.6/etc/hadoop
解释:让Oozie引用Hadoop的配置文件

6. 在Mysql中创建Oozie的数据库
进入Mysql并创建oozie数据库:
$ mysql -uroot -p123456
mysql> create database oozie;

7、初始化Oozie
1)上传Oozie目录下的yarn.tar.gz文件到HDFS:
oozie会自动解压上传到hdfs
[kris@hadoop101 oozie-4.0.0-cdh5.3.6]$ bin/oozie-setup.sh sharelib create -fs hdfs://hadoop101:8020 -locallib oozie-sharelib-4.0.0-cdh5.3.6-yarn.tar.gz
执行成功之后,去50070检查对应目录有没有文件生成。
2)创建oozie.sql文件 ;它会生成一个.sql文件,如果不指定它会在tmp那个目录创建.sql文件
[kris@hadoop101 oozie-4.0.0-cdh5.3.6]$ bin/ooziedb.sh create -sqlfile oozie.sql -run
  setting CATALINA_OPTS="$CATALINA_OPTS -Xmx1024m"
3)打包项目,生成war包
[kris@hadoop101 oozie-4.0.0-cdh5.3.6]$ bin/oozie-setup.sh prepare-war
  setting CATALINA_OPTS="$CATALINA_OPTS -Xmx1024m"
New Oozie WAR file with added 'ExtJS library, JARs' at /opt/module/oozie-4.0.0-cdh5.3.6/oozie-server/webapps/oozie.war
4) Oozie的启动与关闭 [kris@hadoop101 oozie-4.0.0-cdh5.3.6]$ bin/oozied.sh start [kris@hadoop101 oozie-4.0.0-cdh5.3.6]$ bin/oozied.sh stop

访问Oozie的Web页面

http://hadoop101:11000/oozie/

如果没有oozied.sh 起来,把oozied.pid文件给删掉,重新起下;
[kris@hadoop101 temp]$ pwd
/opt/module/oozie-4.0.0-cdh5.3.6/oozie-server/temp
[kris@hadoop101 temp]$ ll
总用量 8
drwxrwxr-x. 3 kris kris 4096 3月   2 10:52 oozie-kris4439005802986311795.dir
-rw-rw-r--. 1 kris kris    5 3月   2 10:52 oozie.pid
-rwxr-xr-x. 1 kris kris    0 11月 14 2014 safeToDelete.tmp

 

Oozie的使用

Oozie调度shell脚本

1)创建工作目录
[kris@hadoop101 oozie-4.0.0-cdh5.3.6]$ mkdir -p oozie-apps/shell
2)在oozie-apps/shell目录下创建两个文件——job.properties和workflow.xml文件
[kris@hadoop101 shell]$ touch workflow.xml
[kris@hadoop101 shell]$ touch job.properties
3)编辑job.properties和workflow.xml文件

job.properties

#HDFS地址
nameNode=hdfs://hadoop101:8020
#ResourceManager地址
jobTracker=hadoop102:8032
#队列名称
queueName=default
examplesRoot=oozie-apps
oozie.wf.application.path=${nameNode}/user/${user.name}/${examplesRoot}/shell
View Code

workflow.xml

<workflow-app xmlns="uri:oozie:workflow:0.4" name="shell-wf">
<!--开始节点-->
<start to="shell-node"/>
<!--动作节点-->
<action name="shell-node">
    <!--shell动作-->
    <shell xmlns="uri:oozie:shell-action:0.2">
        <job-tracker>${jobTracker}</job-tracker>
        <name-node>${nameNode}</name-node>
        <configuration>
            <property>
                <name>mapred.job.queue.name</name>
                <value>${queueName}</value>
            </property>
        </configuration>
        <!--要执行的脚本-->
        <exec>mkdir</exec>
        <argument>/opt/module/d</argument>
        <capture-output/>
    </shell>
    <ok to="end"/>
    <error to="fail"/>
</action>
<!--kill节点-->
<kill name="fail">
    <message>Shell action failed, error message[${wf:errorMessage(wf:lastErrorNode())}]</message>
</kill>
<!--结束节点-->
<end name="end"/>
</workflow-app>
View Code

 job-tracker是hadoop1.x 中ResourceManager

oozie.coord.application.path:${nameNode}/user/${user.name}/${examplesRoot}/shell 这个路径是这两个配置文件要上传到HDFS上的路径

              hdfs://hadoop101:8020/user/kris/oozie-apps/shell

[kris@hadoop101 shell]$ ll
总用量 8
-rw-rw-r--. 1 kris kris 226 3月   2 18:36 job.properties
-rw-rw-r--. 1 kris kris 900 3月   2 18:37 workflow.xml

[kris@hadoop101 oozie-4.0.0-cdh5.3.6]$ /opt/module/cdh/hadoop-2.5.0-cdh5.3.6/bin/hadoop fs -put oozie-apps/ /user/kris
19/03/02 18:41:08 WARN util.NativeCodeLoader: Unable to load native-hadoop library for your platform... using builtin-java classes where applicable
[kris@hadoop101 oozie-4.0.0-cdh5.3.6]$ bin/oozie job -oozie http://hadoop101:11000/oozie -config oozie-apps/shell/job.properties -run
job: 0000000-190302183330904-oozie-kris-W

DAG有向无环图,没有形成死循环
  start

  ↓
  action → kill

  ↓
  end

颜色是绿色才代表成功,否则就是出错了;

杀死某个任务
[kris@hadoop101 oozie-4.0.0-cdh5.3.6]$ bin/oozie job -oozie http://hadoop101:11000/oozie -kill 0000000-190302183330904-oozie-kris-W

 

Oozie逻辑调度执行多个Job| 两个工作单元

job.properties

nameNode=hdfs://hadoop101:8020
jobTracker=hadoop102:8032
queueName=default
examplesRoot=oozie-apps
oozie.wf.application.path=${nameNode}/user/${user.name}/${examplesRoot}/shells
View Code

workflow.xml

<workflow-app xmlns="uri:oozie:workflow:0.4" name="shell-wf">
    <start to="p1-shell-node"/>
    <action name="p1-shell-node">
        <shell xmlns="uri:oozie:shell-action:0.2">
            <job-tracker>${jobTracker}</job-tracker>
            <name-node>${nameNode}</name-node>
            <configuration>
                <property>
                    <name>mapred.job.queue.name</name>
                    <value>${queueName}</value>
                </property>
            </configuration>
            <exec>mkdir</exec>
            <argument>/opt/module/d1</argument>
            <capture-output/>
        </shell>
        <ok to="p2-shell-node"/>
        <error to="fail"/>
    </action>

    <action name="p2-shell-node">
        <shell xmlns="uri:oozie:shell-action:0.2">
            <job-tracker>${jobTracker}</job-tracker>
            <name-node>${nameNode}</name-node>
            <configuration>
                <property>
                    <name>mapred.job.queue.name</name>
                    <value>${queueName}</value>
                </property>
            </configuration>
            <exec>mkdir</exec>
            <argument>/opt/module/d2</argument>
            <capture-output/>
        </shell>
        <ok to="end"/>
        <error to="fail"/>
    </action>
    <kill name="fail">
        <message>Shell action failed, error message[${wf:errorMessage(wf:lastErrorNode())}]</message>
    </kill>
    <end name="end"/>
</workflow-app>
View Code

 

上传任务配置
[kris@hadoop101 hadoop-2.5.0-cdh5.3.6]$ /opt/module/cdh/hadoop-2.5.0-cdh5.3.6/bin/hadoop fs -put oozie-apps/shells/ /user/kris/oozie-apps19/03/02 19:06:04 WARN util.NativeCodeLoader: Unable to load native-hadoop library for your platform... using builtin-java classes where applicable
执行任务
[kris@hadoop101 oozie-4.0.0-cdh5.3.6]$ bin/oozie job -oozie http://hadoop101:11000/oozie -config oozie-apps/shells/job.properties -run
job: 0000001-190302183330904-oozie-kris-W

Oozie调用有分支的任务| 4个工作单元

fock和join的用法见官网:http://oozie.apache.org/docs/4.0.0/WorkflowFunctionalSpec.html#a3.1.5_Fork_and_Join_Control_Nodes 

             ---to-->p2-shell-node 

p1-shell-node --to-->forking                                 ---它俩 --join---to -->p4-shell-node

              ---to-->p3-shell-node

 

job.properties

nameNode=hdfs://hadoop101:8020
jobTracker=hadoop102:8032
queueName=default
examplesRoot=oozie-apps
oozie.wf.application.path=${nameNode}/user/${user.name}/${examplesRoot}/xshell
View Code

workflow.xml

<workflow-app xmlns="uri:oozie:workflow:0.4" name="shell-wf">
    <start to="p1-shell-node"/>
    <action name="p1-shell-node">
        <shell xmlns="uri:oozie:shell-action:0.2">
            <job-tracker>${jobTracker}</job-tracker>
            <name-node>${nameNode}</name-node>
            <configuration>
                <property>
                    <name>mapred.job.queue.name</name>
                    <value>${queueName}</value>
                </property>
            </configuration>
            <exec>mkdir</exec>
            <argument>/opt/module/d1</argument>
            <capture-output/>
        </shell>
        <ok to="forking"/>
        <error to="fail"/>
    </action>

    <action name="p2-shell-node">
        <shell xmlns="uri:oozie:shell-action:0.2">
            <job-tracker>${jobTracker}</job-tracker>
            <name-node>${nameNode}</name-node>
            <configuration>
                <property>
                    <name>mapred.job.queue.name</name>
                    <value>${queueName}</value>
                </property>
            </configuration>
            <exec>mkdir</exec>
            <argument>/opt/module/d2</argument>
            <capture-output/>
        </shell>
        <ok to="joining"/>
        <error to="fail"/>
    </action>
    
    <action name="p3-shell-node">
        <shell xmlns="uri:oozie:shell-action:0.2">
            <job-tracker>${jobTracker}</job-tracker>
            <name-node>${nameNode}</name-node>
            <configuration>
                <property>
                    <name>mapred.job.queue.name</name>
                    <value>${queueName}</value>
                </property>
            </configuration>
            <exec>mkdir</exec>
            <argument>/opt/module/d3</argument>
            <capture-output/>
        </shell>
        <ok to="joining"/>
        <error to="fail"/>
    </action>
    <action name="p4-shell-node">
        <shell xmlns="uri:oozie:shell-action:0.2">
            <job-tracker>${jobTracker}</job-tracker>
            <name-node>${nameNode}</name-node>
            <configuration>
                <property>
                    <name>mapred.job.queue.name</name>
                    <value>${queueName}</value>
                </property>
            </configuration>
            <exec>mkdir</exec>
            <argument>/opt/module/d4</argument>
            <capture-output/>
        </shell>
        <ok to="end"/>
        <error to="fail"/>
    </action>
    
    <fork name="forking">
        <path start="p2-shell-node"/>
        <path start="p3-shell-node"/>
    </fork>
    <join name="joining" to="p4-shell-node"/>
    
    
    <kill name="fail">
        <message>Shell action failed, error message[${wf:errorMessage(wf:lastErrorNode())}]</message>
    </kill>
    <end name="end"/>
</workflow-app>
View Code
上传
[kris@hadoop101 oozie-4.0.0-cdh5.3.6]$ /opt/module/cdh/hadoop-2.5.0-cdh5.3.6/bin/hadoop fs -put oozie-apps/xshell/ /user/kris/oozie-apps
19/03/03 14:18:33 WARN util.NativeCodeLoader: Unable to load native-hadoop library for your platform... using builtin-java classes where applicable
执行任务
[kris@hadoop101 oozie-4.0.0-cdh5.3.6]$ bin/oozie job -oozie http://hadoop101:11000/oozie -config oozie-apps/xshell/job.properties -run
job: 0000011-190303122200913-oozie-kris-W

 

Oozie调度MapReduce任务

oozie.examples.tar.gz包:即所有的案例:

[kris@hadoop101 apps]$ ll
总用量 88
drwxr-xr-x. 3 kris kris 4096 3月   2 19:19 aggregator
drwxr-xr-x. 2 kris kris 4096 3月   2 19:19 bundle
drwxr-xr-x. 2 kris kris 4096 3月   2 19:19 cron ##新的
drwxr-xr-x. 2 kris kris 4096 3月   2 19:19 cron-schedule  ##配置定时的,适合linux
drwxr-xr-x. 3 kris kris 4096 3月   2 19:19 custom-main
drwxr-xr-x. 3 kris kris 4096 3月   2 19:19 datelist-java-main
drwxr-xr-x. 3 kris kris 4096 3月   2 19:19 demo
drwxr-xr-x. 2 kris kris 4096 3月   2 19:19 distcp
drwxr-xr-x. 3 kris kris 4096 3月   2 19:19 hadoop-el
drwxr-xr-x. 2 kris kris 4096 3月   2 19:19 hcatalog
drwxr-xr-x. 2 kris kris 4096 3月   2 19:19 hive
drwxr-xr-x. 3 kris kris 4096 3月   2 19:19 java-main #java类
drwxr-xr-x. 3 kris kris 4096 3月   2 19:19 map-reduce #
drwxr-xr-x. 2 kris kris 4096 3月   2 19:19 no-op
drwxr-xr-x. 2 kris kris 4096 3月   2 19:19 pig
drwxr-xr-x. 2 kris kris 4096 3月   2 19:19 shell #
drwxr-xr-x. 2 kris kris 4096 3月   2 19:19 sla
drwxr-xr-x. 2 kris kris 4096 3月   2 19:19 sqoop #
drwxr-xr-x. 2 kris kris 4096 3月   2 19:19 sqoop-freeform
drwxr-xr-x. 2 kris kris 4096 3月   2 19:19 ssh
drwxr-xr-x. 2 kris kris 4096 3月   2 19:19 streaming
drwxr-xr-x. 2 kris kris 4096 3月   2 19:19 subwf
[kris@hadoop101 apps]$ pwd
/opt/module/oozie-4.0.0-cdh5.3.6/examples/apps
[kris@hadoop101 map-reduce]$ ll
总用量 20
-rw-r--r--. 1 kris kris 1012 7月  29 2015 job.properties
-rw-r--r--. 1 kris kris 1028 7月  29 2015 job-with-config-class.properties #带有配置类的;之前写的那三个mapreduce类,Driver是输入输出路径以及k,v格式类型;roozie调用MR时有两种配置:
一种是专门写一个配置类,oozie提供好了接口到时去继承它就可以,实现里边的方法;
另外一种是直接在workflow.xml文件中配置;我们采用第二种方式; drwxrwxr
-x. 2 kris kris 4096 3月 2 19:19 lib #放jar包; -rw-r--r--. 1 kris kris 2274 7月 29 2015 workflow-with-config-class.xml -rw-r--r--. 1 kris kris 2559 7月 29 2015 workflow.xml

 

1) 拷贝官方模板到oozie-apps
 [kris@hadoop101 oozie-4.0.0-cdh5.3.6]$ cp -r /opt/module/oozie-4.0.0-cdh5.3.6/examples/apps/map-reduce/ oozie-apps/
    
[kris@hadoop101 hadoop-2.5.0-cdh5.3.6]$ bin/hadoop fs -mkdir /input
19/03/02 19:34:02 WARN util.NativeCodeLoader: Unable to load native-hadoop library for your platform... using builtin-java classes where applicable
2) 测试一下wordcount在yarn中的运行
[kris@hadoop101 oozie-4.0.0-cdh5.3.6]$ /opt/module/cdh/hadoop-2.5.0-cdh5.3.6/bin/yarn jar /opt/module/cdh/hadoop-2.5.0-cdh5.3.6/share/hadoop/mapreduce/hadoop-mapreduce-examples-2.5.0-cdh5.3.6.jar wordcount /input/ /output/
19/03/02 19:34:32 INFO mapreduce.Job:  map 0% reduce 100%
19/03/02 19:34:33 INFO mapreduce.Job: Job job_1551522372459_0005 completed successfully
19/03/02 19:34:33 INFO mapreduce.Job: Counters: 38
########下面是另外一种:
  创建输入文件wordcount.txt [kris@hadoop101 hadoop
-2.5.0-cdh5.3.6]$ vim wordcount.txt java java java kafka kafka oozie   把它上传到hdfs上 [kris@hadoop101 hadoop-2.5.0-cdh5.3.6]$ bin/hadoop fs -put wordcount.txt /   执行下官方提供的wordcount的jar包 [kris@hadoop101 hadoop-2.5.0-cdh5.3.6]$ bin/yarn jar share/hadoop/mapreduce/hadoop-mapreduce-examples-2.5.0-cdh5.3.6.jar wordcount /wordcount.txt /output
  把官方案例的wordcount的jar包放在lib下面 [kris@hadoop101 lib]$ ll 总用量
28 -rw-r--r--. 1 kris kris 24707 3月 3 14:49 oozie-examples-4.0.0-cdh5.3.6.jar [kris@hadoop101 lib]$ rm -rf oozie-examples-4.0.0-cdh5.3.6.jar [kris@hadoop101 lib]$ cp /opt/module/cdh/hadoop-2.5.0-cdh5.3.6/share/hadoop/mapreduce/hadoop-mapreduce-examples-2.5.0-cdh5.3.6.jar ./ [kris@hadoop101 lib]$ ll 总用量 272 -rw-r--r--. 1 kris kris 275958 3月 3 14:54 hadoop-mapreduce-examples-2.5.0-cdh5.3.6.jar [kris@hadoop101 lib]$

 

3) 配置map-reduce任务的job.properties以及workflow.xml

 

job.properties

nameNode=hdfs://hadoop101:8020
jobTracker=hadoop102:8032
queueName=default
examplesRoot=oozie-apps

oozie.wf.application.path=${nameNode}/user/${user.name}/${examplesRoot}/map-reduce/workflow.xml
outputDir=map-reduce
View Code

workflow.xml

<workflow-app xmlns="uri:oozie:workflow:0.2" name="map-reduce-wf">
    <start to="mr-node"/>
    <action name="mr-node">
        <map-reduce>
            <job-tracker>${jobTracker}</job-tracker>
            <name-node>${nameNode}</name-node>
            <prepare>
                <delete path="${nameNode}/output/"/>
            </prepare>
            <configuration>
                <property>
                    <name>mapred.job.queue.name</name>
                    <value>${queueName}</value>
                </property>
                <!-- 配置调度MR任务时,使用新的API -->
                <property>
                    <name>mapred.mapper.new-api</name>
                    <value>true</value>
                </property>

                <property>
                    <name>mapred.reducer.new-api</name>
                    <value>true</value>
                </property>

                <!-- 指定Job Key输出类型 -->
                <property>
                    <name>mapreduce.job.output.key.class</name>
                    <value>org.apache.hadoop.io.Text</value>
                </property>

                <!-- 指定Job Value输出类型 -->
                <property>
                    <name>mapreduce.job.output.value.class</name>
                    <value>org.apache.hadoop.io.IntWritable</value>
                </property>

                <!-- 指定输入路径 -->
                <property>
                    <name>mapred.input.dir</name>
                    <value>/wordcount.txt</value>
                </property>

                <!-- 指定输出路径 -->
                <property>
                    <name>mapred.output.dir</name>
                    <value>/output/</value>
                </property>

                <!-- 指定Map类 -->
                <property>
                    <name>mapreduce.job.map.class</name>
                    <value>org.apache.hadoop.examples.WordCount$TokenizerMapper</value>
                </property>

                <!-- 指定Reduce类 -->
                <property>
                    <name>mapreduce.job.reduce.class</name>
                    <value>org.apache.hadoop.examples.WordCount$IntSumReducer</value>
                </property>
            </configuration>
        </map-reduce>
        <ok to="end"/>
        <error to="fail"/>
    </action>
    <kill name="fail">
        <message>Map/Reduce failed, error message[${wf:errorMessage(wf:lastErrorNode())}]</message>
    </kill>
    <end name="end"/>
</workflow-app>
View Code

 <delete path="${nameNode}/output/"/> ##跑在mapreduce中输入路径不能存在;即把output输入文件给删掉

#mapred是不对的,老版的;可以去官网有新旧的对比;也可以直接看wordcount的源码看设置的什么属性

如果map端输出的k v和最终输出的k v类型是一样的,配置一个就可以了

#mapreduce的个数不指定,让它自己决定;看它的源码反编译,jar包反编译就可以得出它的value

                <property>
                    <name>mapred.map.tasks</name> ##可以不指定
                    <value>1</value>
                </property>

5) 上传配置好的文件夹到HDFS
[kris@hadoop101 oozie-4.0.0-cdh5.3.6]$ /opt/module/cdh/hadoop-2.5.0-cdh5.3.6/bin/hdfs dfs -put oozie-apps/map-reduce/ /user/kris/oozie-apps
6) 执行任务
[kris@hadoop101 oozie-4.0.0-cdh5.3.6]$ bin/oozie job -oozie http://hadoop101:11000/oozie -config oozie-apps/map-reduce/job.properties -run
job: 0000000-190302224616759-oozie-kris-W

 

 

Oozie定时任务/循环任务

目标:Coordinator周期性调度任务

分步实现:

1)  配置Linux时区以及时间服务器

2)  检查系统当前时区:

# date -R

 

[kris@hadoop101 hadoop-2.5.0-cdh5.3.6]$ date -R
Sun, 03 Mar 2019 12:58:37 +0800
[kris@hadoop102 module]$ date -R
Sun, 03 Mar 2019 12:58:02 +0800
[kris@hadoop103 ~]$ date -R
Sun, 03 Mar 2019 12:58:56 +0800

 

  配置oozie-site.xml文件;  注:该属性去oozie-default.xml中找到即可
[kris@hadoop101 conf]$ vim oozie-site.xml ;将默认的UTC改成GMT+0800
<property>
        <name>oozie.processing.timezone</name>
        <value>GMT+0800</value>
        <description>
            Oozie server timezone. Valid values are UTC and GMT(+/-)####, for example 'GMT+0530' would be India
            timezone. All dates parsed and genered dates by Oozie Coordinator/Bundle will be done in the specified
            timezone. The default value of 'UTC' should not be changed under normal circumtances. If for any reason
            is changed, note that GMT(+/-)#### timezones do not observe DST changes.
        </description>
    </property>

 修改js框架中的关于时间设置的代码
function getTimeZone() {
    Ext.state.Manager.setProvider(new Ext.state.CookieProvider());
    return Ext.state.Manager.get("TimezoneId","GMT+0800");
}
重启oozie服务,并重启浏览器(一定要注意清除缓存)
[kris@hadoop101 oozie-4.0.0-cdh5.3.6]$ bin/oozied.sh stop
[kris@hadoop101 oozie-4.0.0-cdh5.3.6]$ bin/oozied.sh start

拷贝官方模板配置定时任务\
[kris@hadoop101 oozie-4.0.0-cdh5.3.6]$ cp -r examples/apps/cron oozie-apps/

修改模板job.properties和coordinator.xml以及workflow.xml

 job.properties

nameNode=hdfs://hadoop101:8020
jobTracker=hadoop102:8032
queueName=default
examplesRoot=oozie-apps

oozie.coord.application.path=${nameNode}/user/${user.name}/${examplesRoot}/cron
#start:必须设置为未来时间,否则任务失败
start=2019-03-03T12:33+0800
end=2019-03-03T12:53+0800
workflowAppUri=${nameNode}/user/${user.name}/${examplesRoot}/cron
EXEC=p1.sh
View Code

coordinator.xml

<coordinator-app name="cron-coord" frequency="${coord:minutes(5)}" start="${start}" end="${end}" timezone="GMT+0800" xmlns="uri:oozie:coordinator:0.2">
<action>
    <workflow>
        <app-path>${workflowAppUri}</app-path>
        <configuration>
            <property>
                <name>jobTracker</name>
                <value>${jobTracker}</value>
            </property>
            <property>
                <name>nameNode</name>
                <value>${nameNode}</value>
            </property>
            <property>
                <name>queueName</name>
                <value>${queueName}</value>
            </property>
        </configuration>
    </workflow>
</action>
</coordinator-app>
View Code

workflow.xml

<workflow-app xmlns="uri:oozie:workflow:0.5" name="one-op-wf">
<start to="shell-node"/>
  <action name="shell-node">
      <shell xmlns="uri:oozie:shell-action:0.2">
          <job-tracker>${jobTracker}</job-tracker>
          <name-node>${nameNode}</name-node>
          <configuration>
              <property>
                  <name>mapred.job.queue.name</name>
                  <value>${queueName}</value>
              </property>
          </configuration>
          <exec>${EXEC}</exec>
          <file>/user/kris/oozie-apps/cron/${EXEC}#${EXEC}</file>
          <capture-output/>
      </shell>
      <ok to="end"/>
      <error to="fail"/>
  </action>
<kill name="fail">
    <message>Shell action failed, error message[${wf:errorMessage(wf:lastErrorNode())}]</message>
</kill>
<end name="end"/>
</workflow-app>
View Code

其中 #${EXEC}是给起别名; 需要自定义配置的:frequency start end timezone;定时调度的workflow:路径hdfs上的

workflowAppUri 这个路径是放workflow.xml文件和job.properties的HDFS上的路径

<!--要执行的脚本的写法-->

<exec>p1.sh</exec>
  <file>/opt/module/oozie-apps/cron/p1.sh</file> #hdfs上的路径
<capture-output/>
p1.sh也可以自己创建lib目录下就不需写路径了

yarn通过container收集日志;可以查看它的log日志

[kris@hadoop101 cron]$ vim p1.sh  #这里要写绝对路径
#!/bin/bash
date >> /opt/module/p1.sh
上传配置
[kris@hadoop101 oozie-4.0.0-cdh5.3.6]$ /opt/module/cdh/hadoop-2.5.0-cdh5.3.6/bin/hadoop fs -put oozie-apps/cron/ /user/kris/oozie-apps
启动任务
[kris@hadoop101 oozie-4.0.0-cdh5.3.6]$ bin/oozie job -oozie http://hadoop101:11000/oozie -config oozie-apps/cron/job.properties -run
job: 0000002-190303122200913-oozie-kris-C

可以看到
[kris@hadoop102 module]$ cat p1.sh 
2019年 03月 03日 星期日 12:32:14 CST
2019年 03月 03日 星期日 12:33:09 CST
2019年 03月 03日 星期日 12:37:07 CST

 

hive的配置

[kris@hadoop101 hive]$ ll
总用量 12
-rw-r--r--. 1 kris kris 1000 7月  29 2015 job.properties
-rw-r--r--. 1 kris kris  939 7月  29 2015 script.q
-rw-r--r--. 1 kris kris 2003 7月  29 2015 workflow.xml
[kris@hadoop101 hive]$ pwd
/opt/module/oozie-4.0.0-cdh5.3.6/examples/apps/hive
[kris@hadoop101 hive]$ vim script.q 
CREATE EXTERNAL TABLE test (a INT) STORED AS TEXTFILE LOCATION '${INPUT}';
INSERT OVERWRITE DIRECTORY '${OUTPUT}' SELECT * FROM test;
[kris@hadoop101 hive]$ vim workflow.xml
<script>script.q</script>...

 

 

posted @ 2019-03-03 15:15  kris12  阅读(1035)  评论(0编辑  收藏  举报
levels of contents