Project

General

Profile

1 56002 enrico.ott
<RESOURCE_PROFILE>
2
	<HEADER>
3
		<RESOURCE_IDENTIFIER
4
				value="249e40c2-6420-4207-b40a-e1236f77f1fc_SGFkb29wSm9iQ29uZmlndXJhdGlvbkRTUmVzb3VyY2VzL0hhZG9vcEpvYkNvbmZpZ3VyYXRpb25EU1Jlc291cmNlVHlwZQ=="/>
5
		<RESOURCE_TYPE value="HadoopJobConfigurationDSResourceType"/>
6
		<RESOURCE_KIND value="HadoopJobConfigurationDSResources"/>
7
		<RESOURCE_URI value=""/>
8
		<DATE_OF_CREATION value="2019-05-29T12:00:00"/>
9
	</HEADER>
10
	<BODY>
11
		<HADOOP_JOB name="importOrcidJob" type="mapreduce">
12
			<DESCRIPTION>map reduce job that import the Orcid works (no doi) into actions</DESCRIPTION>
13
			<STATIC_CONFIGURATION>
14
15
				<!-- I/O FORMAT -->
16
				<PROPERTY key="mapreduce.inputformat.class" value="org.apache.hadoop.mapreduce.lib.input.TextInputFormat"/>
17
				<PROPERTY key="mapreduce.outputformat.class" value="org.apache.hadoop.mapreduce.lib.output.SequenceFileOutputFormat"/>
18
19
				<!-- MAPPER -->
20
				<PROPERTY key="mapreduce.map.class" value="eu.dnetlib.data.mapreduce.hbase.dataimport.OrcidImportMapper"/>
21
				<PROPERTY key="mapred.mapoutput.key.class" value="org.apache.hadoop.io.LongWritable"/>
22
				<PROPERTY key="mapred.mapoutput.value.class" value="org.apache.hadoop.io.Text"/>
23
24
				<PROPERTY key="mapred.output.key.class" value="org.apache.hadoop.io.Text"/>
25
				<PROPERTY key="mapred.output.value.class" value="org.apache.hadoop.io.Text"/>
26
27
				<!-- MISC -->
28
				<PROPERTY key="mapred.compress.map.output" value="false"/>
29
				<PROPERTY key="mapred.map.tasks.speculative.execution" value="false"/>
30
				<PROPERTY key="mapreduce.map.speculative" value="false"/>
31
32
				<PROPERTY key="mapred.reduce.tasks" value="0"/>
33
				<PROPERTY key="dfs.blocksize" value="256M"/>
34
35
                <!-- Orcid Mapper Properties -->
36
                <PROPERTY key="setName" value="orcidworks-no-doi"/>
37
				<PROPERTY key="agentId" value="dnet"/>
38
				<PROPERTY key="agentName" value="D-Net"/>
39
				<PROPERTY key="invisible" value="true"/>
40
41
				<!-- <PROPERTY key="user.name" value="dnet" /> -->
42
43
				<!--  	Uncomment to override the default lib path -->
44
				<!--  PROPERTY key="job.lib" value="/lib/dnet/snapshots/dnet-mapreduce-jobs-orcid_beta.jar"/ -->
45
			</STATIC_CONFIGURATION>
46
			<JOB_INTERFACE>
47
				<PARAM name="mapred.input.dir" required="true" description="source sequence file on hdfs"/>
48
				<PARAM name="mapred.output.dir" required="true" description="target sequence file on hdfs"/>
49
			</JOB_INTERFACE>
50
			<SCAN>
51
				<FILTERS/>
52
				<FAMILIES/>
53
			</SCAN>
54
		</HADOOP_JOB>
55
		<STATUS/>
56
		<SECURITY_PARAMETERS>SECURITY_PARAMETERS</SECURITY_PARAMETERS>
57
	</BODY>
58
</RESOURCE_PROFILE>