Project

General

Profile

1
<RESOURCE_PROFILE>
2
    <HEADER>
3
        <RESOURCE_IDENTIFIER value="fa0e188c-aec4-4877-93b4-43e3c5acae1d_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="dedupIndexFeedJob" type="mapreduce">
11
 			<DESCRIPTION>map reduce job that updates the dedup index</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.apache.hadoop.mapreduce.lib.output.SequenceFileOutputFormat" />				
17
        	
18
        		<!-- MAPPER -->
19
				<PROPERTY key="mapreduce.map.class" value="eu.dnetlib.data.mapreduce.hbase.index.DedupIndexFeedMapper" />
20
				<PROPERTY key="mapred.mapoutput.key.class" value="org.apache.hadoop.io.Text" />
21
				<PROPERTY key="mapred.mapoutput.value.class" value="org.apache.hadoop.io.Text" />
22
				
23
				<!-- JOB GLOBAL -->		
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.map.tasks.speculative.execution" value="false" />	
29
				<PROPERTY key="mapreduce.map.speculative" value="false" />		
30
				<PROPERTY key="mapred.reduce.tasks" value="0" />
31
				<PROPERTY key="mapred.fairscheduler.pool" value="solr"/>
32
				<!-- <PROPERTY key="user.name" value="dnet" /> -->
33
				
34
		<!--  	Uncomment to override the default lib path -->			
35
		<!--	<PROPERTY key="job.lib" value="/user/dnet/dnet-mapreduce-jobs-0.0.2-SNAPSHOT-jar-with-dependencies.jar"/> -->
36
        	</STATIC_CONFIGURATION>
37
        	<JOB_INTERFACE>
38
        		<PARAM name="hbase.mapred.inputtable" required="true" description="source hbase table" />
39
	       		<PARAM name="hbase.mapreduce.inputtable" required="true" description="source hbase table" />
40
        		<PARAM name="mapred.output.dir" required="true" description="destination path on hdfs for rotten index xml records" />
41
        		
42
        		<PARAM name="index.solr.url" required="false" description="url used to instantiate the solr client" /> 
43
       			<PARAM name="index.solr.collection" required="true" description="target solr collection to be fed" />
44

    
45
       			<PARAM name="id" required="true" description="index DS id" />
46
				<PARAM name="index.shutdown.wait.time" required="true" description="wait time before shut down the solr client pool" />
47
       			<PARAM name="index.buffer.flush.threshold" required="true" description="indexing buffer flush threshold" />
48
       			<PARAM name="index.feed.timestamp" required="true" description="timestamp used as ds_version" />
49
				<PARAM name="index.solr.sim.mode" required="true" description="boolean value, allows to run this job in simulation mode" />
50
				<PARAM name="index.fields" required="true" description="fields from a given MDFormatDSResourceType" />
51
				<PARAM name="entityType" required="true" description="entity Type name" />
52
        	</JOB_INTERFACE>
53
        	<SCAN>
54
        		<FILTERS operator="MUST_PASS_ALL">
55
        			<FILTER type="prefix" param="entityTypeId" />
56
        		</FILTERS>
57
        		<FAMILIES>
58
        			<FAMILY param="entityType" />
59
        		</FAMILIES>
60
        	</SCAN>
61
        </HADOOP_JOB>
62
        <STATUS>
63
            <LAST_SUBMISSION_DATE value="2001-12-31T12:00:00"/>
64
            <RUNNING_INSTANCES value="0"/>
65
            <CUMULATIVE_RUN value="0" />
66
        </STATUS>
67
        <SECURITY_PARAMETERS>SECURITY_PARAMETERS</SECURITY_PARAMETERS>
68
    </BODY>
69
</RESOURCE_PROFILE>
(25-25/89)