Project

General

Profile

1
<RESOURCE_PROFILE>
2
    <HEADER>
3
        <RESOURCE_IDENTIFIER value="17dd747e-f5f2-45d5-8554-9f70343bfe55_SGFkb29wSm9iQ29uZmlndXJhdGlvbkRTUmVzb3VyY2VzL0hhZG9vcEpvYkNvbmZpZ3VyYXRpb25EU1Jlc291cmNlVHlwZQ=="/>
4
        <RESOURCE_TYPE value="HadoopJobConfigurationDSResourceType"/>
5
        <RESOURCE_KIND value="HadoopJobConfigurationDSResources"/>
6
        <RESOURCE_URI value=""/>
7
        <DATE_OF_CREATION value="2001-12-31T12:00:00"/>
8
    </HEADER>
9
    <BODY>
10
        <HADOOP_JOB name="brokerEnrichmentProjectsJob" type="mapreduce">
11
 			<DESCRIPTION>map reduce job that calculates the enrichment events based on the publications dedup results</DESCRIPTION>
12

    
13
	        <STATIC_CONFIGURATION><!-- I/O FORMAT -->
14
		        <PROPERTY key="mapreduce.inputformat.class" value="org.apache.hadoop.hbase.mapreduce.TableInputFormat"/>
15
		        <PROPERTY key="mapreduce.outputformat.class" value="org.elasticsearch.hadoop.mr.EsOutputFormat"/>
16

    
17
		        <!-- MAPPER -->
18
		        <PROPERTY key="mapreduce.map.class" value="eu.dnetlib.data.mapreduce.hbase.broker.enrich.ProjectEnrichmentMapper"/>
19
		        <PROPERTY key="mapred.mapoutput.key.class" value="org.apache.hadoop.hbase.io.ImmutableBytesWritable"/>
20
		        <PROPERTY key="mapred.mapoutput.value.class" value="org.apache.hadoop.hbase.io.ImmutableBytesWritable"/>
21

    
22
		        <!-- REDUCER -->
23
		        <PROPERTY key="mapreduce.reduce.class" value="eu.dnetlib.data.mapreduce.hbase.broker.enrich.ProjectEnrichmentReducer"/>
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.reduce.tasks.speculative.execution" value="false"/>
29
		        <PROPERTY key="mapred.map.tasks.speculative.execution" value="false"/>
30
		        <PROPERTY key="mapreduce.map.speculative" value="false"/>
31
		        <PROPERTY key="mapreduce.reduce.speculative" value="false"/>
32
		        <PROPERTY key="dfs.blocksize" value="256M"/>
33
		        <PROPERTY key="mapred.reduce.tasks" value="4"/>
34

    
35
		        <!-- ES -->
36
		        <PROPERTY key="es.nodes" value="ip-90-147-167-137.ct1.garrservices.it:9200,ip-90-147-167-126.ct1.garrservices.it:9200,ip-90-147-167-13.ct1.garrservices.it:9200,ip-90-147-167-125.ct1.garrservices.it:9200"/>
37
		        <PROPERTY key="es.nodes.resolve.hostname" value="false"/>
38
		        <PROPERTY key="es.batch.write.retry.count " value="10"/>
39
		        <PROPERTY key="es.batch.size.entries " value="500"/>
40
		        <PROPERTY key="es.nodes.wan.only" value="true"/>
41
		        <PROPERTY key="es.resource" value="events/event"/>
42
		        <PROPERTY key="es.input.json" value="yes"/>
43

    
44
		        <!-- BROKER -->
45
		        <PROPERTY key="broker.datasource.id.whitelist" value=""/>
46
		        <PROPERTY key="broker.datasource.id.blacklist" value=""/>
47
		        <PROPERTY key="broker.datasource.untrusted.oa.list" value="opendoar____::8b6dd7db9af49e67306feb59a8bdc52c"/>
48
		        <PROPERTY key="broker.datasource.type.whitelist" value="pubsrepository::unknown,pubsrepository::institutional,pubsrepository::thematic"/><!-- <PROPERTY key="user.name" value="dnet" /> --><!-- <PROPERTY key="user.name" value="dnet" /> --><!--  	Uncomment to override the default lib path -->
49
	        </STATIC_CONFIGURATION>
50
	        <JOB_INTERFACE>
51
		        <PARAM description="source hbase table" name="hbase.mapred.inputtable" required="true"/>
52
		        <PARAM description="source hbase table" name="hbase.mapreduce.inputtable" required="true"/>
53
	        </JOB_INTERFACE>
54
        	<SCAN>
55
        		<FILTERS operator="MUST_PASS_ONE">
56
			        <FILTER type="prefix" value="40" />
57
        			<FILTER type="prefix" value="50" />
58
        		</FILTERS>
59
        		<FAMILIES>
60
	       			<FAMILY value="result" />
61
			        <FAMILY value="project" />
62
        			<FAMILY value="resultResult_dedup_isMergedIn" />
63
			        <FAMILY value="resultProject_outcome_produces" />
64
			        <FAMILY value="resultProject_outcome_isProducedBy" />
65
        		</FAMILIES>
66
        	</SCAN>
67
        </HADOOP_JOB>
68
        <STATUS>
69
            <LAST_SUBMISSION_DATE value="2001-12-31T12:00:00"/>
70
            <RUNNING_INSTANCES value="0"/>
71
            <CUMULATIVE_RUN value="0" />
72
        </STATUS>
73
        <SECURITY_PARAMETERS>SECURITY_PARAMETERS</SECURITY_PARAMETERS>
74
    </BODY>
75
</RESOURCE_PROFILE>
(3-3/56)