Project

General

Profile

1
<RESOURCE_PROFILE>
2
    <HEADER>
3
        <RESOURCE_IDENTIFIER value="03d7af20-63bb-4790-a052-6cdbc1e05fce_SGFkb29wSm9iQ29uZmlndXJhdGlvbkRTUmVzb3VyY2VzL0hhZG9vcEpvYkNvbmZpZ3VyYXRpb25EU1Jlc291cmNlVHlwZQ=="/>
4
        <RESOURCE_TYPE value="HadoopJobConfigurationDSResourceType"/>
5
        <RESOURCE_KIND value="HadoopJobConfigurationDSResources"/>
6
        <RESOURCE_URI value=""/>
7
        <DATE_OF_CREATION value="2015-02-09T12:00:00"/>
8
    </HEADER>
9
    <BODY>
10
        <HADOOP_JOB name="oaiFeedJob" type="mapreduce">
11
 			<DESCRIPTION>map reduce job that feeds the OAI store</DESCRIPTION>
12
        	<STATIC_CONFIGURATION>
13

    
14
				<!-- I/O FORMAT -->
15
				<PROPERTY key="mapreduce.inputformat.class" value="org.apache.hadoop.mapreduce.lib.input.SequenceFileInputFormat" />
16
				<PROPERTY key="mapreduce.outputformat.class" value="org.apache.hadoop.mapreduce.lib.output.NullOutputFormat" />	
17
        	
18
        		<!-- MAPPER -->
19
				<PROPERTY key="mapreduce.map.class" value="eu.dnetlib.data.mapreduce.hbase.oai.OaiFeedMapper" />
20
				<PROPERTY key="mapred.mapoutput.key.class" value="org.apache.hadoop.io.NullWritable" />
21
				<PROPERTY key="mapred.mapoutput.value.class" value="org.apache.hadoop.io.NullWritable" />				
22
				
23
				<!-- JOB GLOBAL -->		
24
<!--                 <PROPERTY key="mapred.output.key.class" value="org.apache.hadoop.io.NullWritable" /> -->
25
<!--                 <PROPERTY key="mapred.output.value.class" value="org.apache.hadoop.io.NullWritable"/> -->
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="services.publisher.oai.datepatterns" value="yyyy-MM-dd,yyyy-MM-dd'T'HH:mm:ssXXX,yyyy-MM-dd'T'HH:mm:ss.SSSX,yyyy-MM-dd'T'HH:mm:ssZ,yyyy-MM-dd'T'HH:mm:ss.SX,yyyy-MM-dd'T'HH:mm:ss" />
32
				<!-- <PROPERTY key="user.name" value="dnet" /> -->
33
				
34
				<!--  	Uncomment to override the default lib path -->			
35
				<PROPERTY key="job.lib" value="/user/dnet/lib/dnet-mapreduce-jobs-assembly-0.0.6.3-SNAPSHOT.jar"/> 
36
        	</STATIC_CONFIGURATION>
37
        	<JOB_INTERFACE>
38
        		<PARAM name="mapred.input.dir" required="true" description="source sequence file on hdfs" />
39
				<PARAM name="services.publisher.oai.collection" required="true" description="target mongodb collection" />
40
				<PARAM name="oaiConfiguration" required="true" description="configuration bean used to guide the OAI feeding" />
41
				<PARAM name="oai.feed.date" required="true" description="timestamp" />				
42
				<PARAM name="services.publisher.oai.host" required="true" description="mongodb host" />
43
				<PARAM name="services.publisher.oai.port" required="true" description="mongodb port" />
44
				<PARAM name="services.publisher.oai.db" required="true" description="mongodb database name" />	
45
				<PARAM name="services.publisher.oai.skipDuplicates" required="true" description="skip duplicated records." />	
46
				<PARAM name="services.publisher.oai.duplicateXPath" required="true" description="records with this xpath are identified as duplicates" />																
47
        	</JOB_INTERFACE>
48
        	<SCAN>
49
        		<FILTERS />
50
        		<FAMILIES />
51
        	</SCAN>
52
        </HADOOP_JOB>
53
        <STATUS>
54
            <LAST_SUBMISSION_DATE value="2001-12-31T12:00:00"/>
55
            <RUNNING_INSTANCES value="0"/>
56
            <CUMULATIVE_RUN value="0" />
57
        </STATUS>
58
        <SECURITY_PARAMETERS>SECURITY_PARAMETERS</SECURITY_PARAMETERS>
59
    </BODY>
60
</RESOURCE_PROFILE>
(63-63/82)