D-Net

<workflow-app xmlns="uri:oozie:workflow:0.3"

	name="test-javamapreduce_cloner_with_multiple_output_without_reducer_with_empty_input">

	<start to="producer"/>

	<action name="producer">

		<java>

			<job-tracker>${jobTracker}</job-tracker>

			<name-node>${nameNode}</name-node>

			<!-- The data generated by this node is deleted in this section -->

			<prepare>

				<delete path="${nameNode}${workingDir}/producer" />

				<mkdir path="${nameNode}${workingDir}/producer" />

			</prepare>

			<configuration>

				<property>

					<name>mapred.job.queue.name</name>

					<value>${queueName}</value>

				</property>

			</configuration>

			<!-- This is simple wrapper for the Java code -->

			<main-class>eu.dnetlib.iis.core.java.ProcessWrapper</main-class>

			<!-- The business Java code that gets to be executed -->

			<arg>eu.dnetlib.iis.core.java.jsonworkflownodes.Producer</arg>

			<!-- Specification of the output ports -->

			<arg>-C{person, 

				eu.dnetlib.iis.core.examples.schemas.documentandauthor.Person,

				eu/dnetlib/iis/core/examples/data/empty.json}</arg>

			<!-- All input and output ports have to be bound to paths in HDFS -->

			<arg>-Operson=${workingDir}/producer/person</arg>

		</java>

		<ok to="cloner" />

		<error to="fail" />

	</action>

	<action name="cloner">

		<map-reduce>

			<job-tracker>${jobTracker}</job-tracker>

			<name-node>${nameNode}</name-node>

			<!-- The data generated by this node in the previous run is 

				deleted in this section -->

			<prepare>

				<delete path="${nameNode}${workingDir}/cloner" />

			</prepare>

			<!-- That's a multiple output MapReduce job, so no need to 

				create cloner directory, since it will be created by 

				MapReduce /> -->

			<configuration>

				<!-- # Standard set of options that stays the same regardless 

					of a concrete definition of map-reduce job -->

				<!-- ## Various options -->

				<!--This property seems to not be needed -->

				<!--<property> <name>mapred.job.queue.name</name> <value>${queueName}</value> 

					</property> -->

				<property>

					<name>mapreduce.inputformat.class</name>

					<value>eu.dnetlib.iis.core.javamapreduce.hack.KeyInputFormat</value>

				</property>

				<!-- The output format is not needed since there is no Reduce phase -->

				<!-- <property>

					<name>mapreduce.outputformat.class</name>

					<value>eu.dnetlib.iis.core.javamapreduce.hack.KeyOutputFormat</value>

				</property>-->

				<property>

					<name>mapred.mapoutput.key.class</name>

					<value>org.apache.avro.mapred.AvroKey</value>

				</property>

				<property>

					<name>mapred.mapoutput.value.class</name>

					<value>org.apache.avro.mapred.AvroValue</value>

				</property>

				<property>

					<name>mapred.output.key.class</name>

					<value>org.apache.avro.mapred.AvroKey</value>

				</property>

				<property>

					<name>mapred.output.value.class</name>

					<value>org.apache.avro.mapred.AvroValue</value>

				</property>

				<property>

					<name>mapred.output.key.comparator.class</name>

					<value>eu.dnetlib.iis.core.javamapreduce.hack.KeyComparator</value>

				</property>

				<property>

					<name>io.serializations</name>

					<value>org.apache.hadoop.io.serializer.WritableSerialization,org.apache.hadoop.io.serializer.avro.AvroSpecificSerialization,org.apache.hadoop.io.serializer.avro.AvroReflectSerialization,org.apache.avro.hadoop.io.AvroSerialization

					</value>

				</property>

				<property>

					<name>mapred.output.value.groupfn.class</name>

					<value>eu.dnetlib.iis.core.javamapreduce.hack.KeyComparator</value>

				</property>

				<property>

					<name>rpc.engine.org.apache.hadoop.hdfs.protocolPB.ClientNamenodeProtocolPB

					</name>

					<value>org.apache.hadoop.ipc.ProtobufRpcEngine</value>

				</property>

				<!-- ## This is required for new MapReduce API usage -->

				<property>

					<name>mapred.mapper.new-api</name>

					<value>true</value>

				</property>

				<property>

					<name>mapred.reducer.new-api</name>

					<value>true</value>

				</property>

				<!-- # Job-specific options -->

				<!-- Since there is no reduce phase, there should be no

				reduce tasks -->

				<property>

                    <name>mapred.reduce.tasks</name>

                    <value>0</value>

                </property>

				<!-- ## Names of all output ports -->

				<property>

					<name>avro.mapreduce.multipleoutputs</name>

					<value>person age</value>

				</property>

				<!-- ## Output classes for all output ports -->

				<property>

					<name>avro.mapreduce.multipleoutputs.namedOutput.person.format

					</name>

					<value>org.apache.avro.mapreduce.AvroKeyOutputFormat</value>

				</property>

				<property>

					<name>avro.mapreduce.multipleoutputs.namedOutput.age.format

					</name>

					<value>org.apache.avro.mapreduce.AvroKeyOutputFormat</value>

				</property>

				<!-- ## Classes of mapper and reducer -->

				<property>

					<name>mapreduce.map.class</name>

					<value>eu.dnetlib.iis.core.examples.javamapreduce.MultipleOutputPersonClonerMapper</value>

				</property>

				<!-- No reducer -->

				<!-- ## Schemas -->

				<!-- ### Schema of the data ingested by the mapper. To be more precise, 

					it's the schema of Avro data passed as template parameter of the AvroKey 

					object passed to mapper. -->

				<property>

					<name>eu.dnetlib.iis.avro.input.class</name>

					<value>eu.dnetlib.iis.core.examples.schemas.documentandauthor.Person</value>

				</property>

				<!-- ### Schemas of the data produced by the mapper -->

				<!-- #### Schema of the key produced by the mapper. 

				To be more precise,	it's the schema of Avro data produced 

				by the mapper and passed forward as	template parameter of 

				AvroKey object. -->

				<!-- As a convention, we're setting "null" values 

				since mapper does not produce any standard data in this example 

				(probably any other valid Avro schema would be OK as well).-->

				<property>

					<name>eu.dnetlib.iis.avro.map.output.key.class</name>

                	<value>org.apache.avro.Schema.Type.NULL</value>

                </property>

				<!-- #### Schema of the value produced by the mapper. To be more precise, 

					it's the schema of Avro data produced by the mapper and passed forward as 

					template parameter of AvroValue object. -->

				<!-- As a convention, we're setting "null" values 

				since mapper does not produce any standard data in this example 

				(probably any other valid Avro schema would be OK as well).-->

				<property>

					<name>eu.dnetlib.iis.avro.map.output.value.class</name>

					<value>org.apache.avro.Schema.Type.NULL</value>

				</property>

				<!-- ### Schema of multiple output ports. -->

				<property>

					<name>eu.dnetlib.iis.avro.multipleoutputs.class.person

					</name>

					<value>eu.dnetlib.iis.core.examples.schemas.documentandauthor.Person</value>

				</property>

				<property>

					<name>eu.dnetlib.iis.avro.multipleoutputs.class.age

					</name>

					<value>eu.dnetlib.iis.core.examples.schemas.documentandauthor.PersonAge</value>

				</property>

				<!-- ## Specification of the input and output data store -->

				<property>

					<name>mapred.input.dir</name>

					<value>${workingDir}/producer/person</value>

				</property>

				<!-- This directory does not correspond to a data store. In fact,

				this directory only contains multiple data stores. It has to

				be set to the name of the workflow node.-->

				<property>

					<name>mapred.output.dir</name>

					<value>${workingDir}/cloner</value>

				</property>

				<!-- ## Workflow node parameters -->

				<property>

					<name>copiesCount</name>

					<value>2</value>

				</property>

			</configuration>

		</map-reduce>

		<ok to="consumer" />

		<error to="fail" />

	</action>

    <action name="consumer">

		<java>

			<job-tracker>${jobTracker}</job-tracker>

			<name-node>${nameNode}</name-node>

			<configuration>

				<property>

					<name>mapred.job.queue.name</name>

					<value>${queueName}</value>

				</property>

			</configuration>

			<!-- This is simple wrapper for the Java code -->

			<main-class>eu.dnetlib.iis.core.java.ProcessWrapper</main-class>

			<!-- The business Java code that gets to be executed -->

			<arg>eu.dnetlib.iis.core.java.jsonworkflownodes.TestingConsumer</arg>

			<!-- Specification of the input ports -->

			<arg>-C{person, 

				eu.dnetlib.iis.core.examples.schemas.documentandauthor.Person,

				eu/dnetlib/iis/core/examples/data/empty.json}</arg>

			<arg>-C{age, 

				eu.dnetlib.iis.core.examples.schemas.documentandauthor.PersonAge,

				eu/dnetlib/iis/core/examples/data/empty.json}</arg>

			<!-- All input and output ports have to be bound to paths in HDFS -->

			<arg>-Iperson=${workingDir}/cloner/person</arg>

			<arg>-Iage=${workingDir}/cloner/age</arg>

		</java>

		<ok to="end" />

		<error to="fail" />

	</action>

	<kill name="fail">

		<message>Unfortunately, the process failed -- error message:

			[${wf:errorMessage(wf:lastErrorNode())}]

		</message>

	</kill>

	<end name="end" />

</workflow-app>