Project

General

Profile

1 40245 claudio.at
<RESOURCE_PROFILE>
2
	<HEADER>
3
		<RESOURCE_IDENTIFIER
4 52964 claudio.at
				value="f96a082f-c10b-40e2-beb3-a1b1ffdb4a08_SGFkb29wSm9iQ29uZmlndXJhdGlvbkRTUmVzb3VyY2VzL0hhZG9vcEpvYkNvbmZpZ3VyYXRpb25EU1Jlc291cmNlVHlwZQ=="/>
5 40245 claudio.at
		<RESOURCE_TYPE value="HadoopJobConfigurationDSResourceType"/>
6
		<RESOURCE_KIND value="HadoopJobConfigurationDSResources"/>
7
		<RESOURCE_URI value=""/>
8
		<DATE_OF_CREATION value="2001-12-31T12:00:00"/>
9
	</HEADER>
10
	<BODY>
11 53780 claudio.at
		<HADOOP_JOB name="importDOIBoostJob" type="mapreduce">
12 52914 sandro.lab
			<DESCRIPTION>map reduce job that import the scholexplorer dump into actions</DESCRIPTION>
13 40245 claudio.at
			<STATIC_CONFIGURATION>
14
15
				<!-- I/O FORMAT -->
16 52936 claudio.at
				<PROPERTY key="mapreduce.inputformat.class" value="org.apache.hadoop.mapreduce.lib.input.TextInputFormat"/>
17 40245 claudio.at
				<PROPERTY key="mapreduce.outputformat.class" value="org.apache.hadoop.mapreduce.lib.output.SequenceFileOutputFormat"/>
18
19
				<!-- MAPPER -->
20 53780 claudio.at
				<PROPERTY key="mapreduce.map.class" value="eu.dnetlib.data.mapreduce.hbase.dataimport.DOIBoostImportMapper"/>
21 40245 claudio.at
				<PROPERTY key="mapred.mapoutput.key.class" value="org.apache.hadoop.io.Text"/>
22 52914 sandro.lab
				<PROPERTY key="mapred.mapoutput.value.class" value="org.apache.hadoop.io.Text"/>
23 40245 claudio.at
24 52936 claudio.at
				<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 53780 claudio.at
				<!-- REDUCER -->
28
				<PROPERTY key="mapreduce.reduce.class" value="eu.dnetlib.data.mapreduce.hbase.dataimport.DOIBoostImportReducer" />
29
				<PROPERTY key="mapred.output.key.class" value="org.apache.hadoop.io.Text" />
30
				<PROPERTY key="mapred.output.value.class" value="org.apache.hadoop.io.Text" />
31
32 40245 claudio.at
				<!-- MISC -->
33
				<PROPERTY key="mapred.compress.map.output" value="false"/>
34 53780 claudio.at
				<PROPERTY key="mapred.reduce.tasks.speculative.execution" value="false"/>
35 40245 claudio.at
				<PROPERTY key="mapred.map.tasks.speculative.execution" value="false"/>
36
				<PROPERTY key="mapreduce.map.speculative" value="false"/>
37 53780 claudio.at
				<PROPERTY key="mapreduce.reduce.speculative" value="false"/>
38 40245 claudio.at
39 54156 claudio.at
				<PROPERTY key="mapred.reduce.tasks" value="32"/>
40 52936 claudio.at
				<PROPERTY key="dfs.blocksize" value="256M"/>
41 52928 sandro.lab
42 53780 claudio.at
				<PROPERTY key="mapred.output.compress" value="true"/>
43
				<PROPERTY key="mapred.output.compression.type" value="BLOCK"/>
44
45 52964 claudio.at
                <!-- Crossref Mapper Properties -->
46
                <PROPERTY key="setName" value="crossref-dump"/>
47
				<PROPERTY key="agentId" value="dnet"/>
48
				<PROPERTY key="agentName" value="D-Net"/>
49
				<PROPERTY key="invisible" value="true"/>
50 52928 sandro.lab
51 40245 claudio.at
				<!-- <PROPERTY key="user.name" value="dnet" /> -->
52
53
				<!--  	Uncomment to override the default lib path -->
54
				<!--	<PROPERTY key="job.lib" value="/user/dnet/dnet-mapreduce-jobs-0.0.2-SNAPSHOT-jar-with-dependencies.jar"/> -->
55
			</STATIC_CONFIGURATION>
56
			<JOB_INTERFACE>
57
				<PARAM name="mapred.input.dir" required="true" description="source sequence file on hdfs"/>
58
				<PARAM name="mapred.output.dir" required="true" description="target sequence file on hdfs"/>
59 53780 claudio.at
				<PARAM name="onlyOrganization" required="false" description="boolean value that declares import mode: affiliations or ALL"/>
60 40245 claudio.at
			</JOB_INTERFACE>
61
			<SCAN>
62
				<FILTERS/>
63
				<FAMILIES/>
64
			</SCAN>
65
		</HADOOP_JOB>
66
		<STATUS>
67
			<LAST_SUBMISSION_DATE value="2001-12-31T12:00:00"/>
68
			<RUNNING_INSTANCES value="0"/>
69
			<CUMULATIVE_RUN value="0"/>
70
		</STATUS>
71
		<SECURITY_PARAMETERS>SECURITY_PARAMETERS</SECURITY_PARAMETERS>
72
	</BODY>
73
</RESOURCE_PROFILE>