How to chain multiple MapReduce jobs in Hadoop

When running MapReduce jobs it is possible to have several MapReduce steps with overall job scenarios means the last reduce output will be used as input for the next map job.

Map1 -> Reduce1 -> Map2 -> Reduce2 -> Map3...

While searching for an answer to my MapReduce job, I stumbled upon several cool new ways to achieve my objective. Here are some of the ways:

 

Using Map/Reduce JobClient.runJob() Library to chain jobs:

 

http://developer.yahoo.com/hadoop/tutorial/module4.html#chaining

You can easily chain jobs together in this fashion by writing multiple driver methods, one for each job. Call the first driver method, which uses JobClient.runJob() to run the job and wait for it to complete. When that job has
completed, then call the next driver method, which creates a new JobConf object referring to different instances of Mapper and Reducer, etc. The first job in the chain should write its output to a path which is then used as the input path for the second job.
This process can be repeated for as many jobs are necessary to arrive at a complete solution to the problem

 

Method 1:

First create the JobConf object "job1" for the first job and set all the parameters with "input" as inputdirectory and "temp" as output directory. Execute this job: JobClient.run(job1).
Immediately below it, create the JobConf object "job2" for the second job and set all the parameters with "temp" as inputdirectory and "output" as output directory. Finally execute second job: JobClient.run(job2).
 

Method 2:

Create two JobConf objects and set all the parameters in them just like (1) except that you don't use JobClient.run.
Then create two Job objects with jobconfs as parameters: Job job1=new Job(jobconf1); Job job2=new Job(jobconf2);
Using the jobControl object, you specify the job dependencies and then run the jobs: JobControl jbcntrl=new JobControl("jbcntrl"); jbcntrl.addJob(job1); jbcntrl.addJob(job2); job2.addDependingJob(job1); jbcntrl.run();
 

Using Oozie which is Hadoop Workflow Service described as below:

https://issues.apache.org/jira/secure/attachment/12400686/hws-v1_0_2009FEB22.pdf

3.1.5 Fork and Join Control Nodes

A fork node splits one path of execution into multiple concurrent paths of execution. A join node waits until every concurrent execution path of a previous fork node arrives to it. fork and join nodes must be used in pairs. The
join node assumes concurrent execution paths are children of the same fork node.

 

The name attribute in the fork node is the name of the workflow fork node. The to attribute in the transition elements in the fork node indicate the name of the workflow node that will be part of the concurrent execution. The
name attribute in the join node is the name of the workflow join node. The to attribute in the transition element in the join node indicates the name of the workflow node that will executed after all  3.1.5 Fork and Join Control Nodes 5concurrent execution
paths of the corresponding fork arrive to the join node.

 

Example:

<hadoop−workflow name="sample−wf">

    ...

    <fork name="forking">

        <transition to="firstparalleljob"/>

        <transition to="secondparalleljob"/>

    </fork>

    <hadoop name="firstparalleljob">

       <job−xml>job1.xml</job−xml>

       <transition name="OK" to="joining"/>

       <transition name="ERROR" to="fail"/>

    </hadoop>

    <hadoop name="secondparalleljob">

        <job−xml>job2.xml</job−xml>

        <transition name="OK" to="joining"/>

        <transition name="ERROR" to="fail"/>

    </hadoop>

    <join name="joining">

        <transition to="nextaction"/>

    </join>

    ...

</hadoop−workflow>

* 注：多个连续的job会产生的中间的文件，在调用完成job之后可以用shell中删除中间的文件