Project

General

Profile

1
<RESOURCE_PROFILE>
2
    <HEADER>
3
        <RESOURCE_IDENTIFIER value="ffcc359c-555a-4d86-9ee2-b8ad212a0185_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="brokerEnrichmentJob" type="mapreduce">
11
 			<DESCRIPTION>map reduce job that calculates the enrichment events based on the publications dedup results</DESCRIPTION>
12
        	<STATIC_CONFIGURATION>
13

    
14
		        <!-- I/O FORMAT -->
15
		        <PROPERTY key="mapreduce.inputformat.class" value="org.apache.hadoop.hbase.mapreduce.TableInputFormat"/>
16
		        <PROPERTY key="mapreduce.outputformat.class" value="org.elasticsearch.hadoop.mr.EsOutputFormat"/>
17

    
18
		        <!-- MAPPER -->
19
		        <PROPERTY key="mapreduce.map.class" value="eu.dnetlib.data.mapreduce.hbase.broker.enrich.EnrichmentMapper"/>
20
		        <PROPERTY key="mapred.mapoutput.key.class" value="org.apache.hadoop.hbase.io.ImmutableBytesWritable"/>
21
		        <PROPERTY key="mapred.mapoutput.value.class" value="org.apache.hadoop.hbase.io.ImmutableBytesWritable"/>
22

    
23
		        <!-- REDUCER -->
24
		        <PROPERTY key="mapreduce.reduce.class" value="eu.dnetlib.data.mapreduce.hbase.broker.enrich.EnrichmentReducer" />
25
		        <PROPERTY key="mapred.output.key.class" value="org.apache.hadoop.io.Text" />
26
		        <PROPERTY key="mapred.output.value.class" value="org.apache.hadoop.io.Text" />
27

    
28

    
29
		        <!-- MISC -->
30
		        <PROPERTY key="mapred.reduce.tasks.speculative.execution" value="false"/>
31
		        <PROPERTY key="mapred.map.tasks.speculative.execution" value="false"/>
32
		        <PROPERTY key="mapreduce.map.speculative" value="false"/>
33
		        <PROPERTY key="mapreduce.reduce.speculative" value="false"/>
34

    
35
		        <PROPERTY key="dfs.blocksize" value="256M"/>
36
		        <PROPERTY key="mapred.reduce.tasks" value="1"/>
37

    
38
		        <!-- ES -->
39
		        <PROPERTY key="es.nodes"
40
		                  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"/>
41
		        <PROPERTY key="es.nodes.resolve.hostname" value="false"/>
42
		        <PROPERTY key="es.nodes.wan.only" value="true"/>
43
		        <PROPERTY key="es.resource" value="events_{infra}/event"/>
44
		        <PROPERTY key="es.input.json" value="yes"/>
45
		        <PROPERTY key="es.mapping.id" value="eventId"/>
46

    
47
		        <!-- BROKER -->
48
		        <PROPERTY key="broker.datasource.id.whitelist" value=""/>
49
		        <PROPERTY key="broker.datasource.id.blacklist" value=""/>
50
		        <PROPERTY key="broker.datasource.untrusted.oa.list" value="opendoar____::8b6dd7db9af49e67306feb59a8bdc52c"/>
51
		        <PROPERTY key="broker.datasource.type.whitelist" value="pubsrepository::unknown,pubsrepository::institutional,pubsrepository::thematic"/>
52

    
53

    
54
		        <!-- <PROPERTY key="user.name" value="dnet" /> -->
55
				<!-- <PROPERTY key="user.name" value="dnet" /> -->
56
				
57
		<!--  	Uncomment to override the default lib path -->			
58
		<!--	<PROPERTY key="job.lib" value="/user/dnet/dnet-mapreduce-jobs-0.0.2-SNAPSHOT-jar-with-dependencies.jar"/> -->
59
        	</STATIC_CONFIGURATION>
60
        	<JOB_INTERFACE>
61
        		<PARAM name="hbase.mapred.inputtable" required="true" description="source hbase table" />
62
        		<PARAM name="hbase.mapreduce.inputtable" required="true" description="source hbase table" />
63
        	</JOB_INTERFACE>
64
        	<SCAN>
65
        		<FILTERS operator="MUST_PASS_ALL">
66
        			<FILTER type="prefix" value="50" />
67
        		</FILTERS>
68
        		<FAMILIES>
69
	       			<FAMILY value="result" />
70
        			<FAMILY value="resultResult_dedup_isMergedIn" />
71
        		</FAMILIES>
72
        	</SCAN>
73
        </HADOOP_JOB>
74
        <STATUS>
75
            <LAST_SUBMISSION_DATE value="2001-12-31T12:00:00"/>
76
            <RUNNING_INSTANCES value="0"/>
77
            <CUMULATIVE_RUN value="0" />
78
        </STATUS>
79
        <SECURITY_PARAMETERS>SECURITY_PARAMETERS</SECURITY_PARAMETERS>
80
    </BODY>
81
</RESOURCE_PROFILE>
(2-2/85)