Project

General

Profile

« Previous | Next » 

Revision 54157

minor changes

View differences:

modules/dnet-openaireplus-profiles/trunk/src/main/resources/eu/dnetlib/test/profiles/HadoopJobConfigurationDSResources/HadoopJobConfigurationDSResourceType/invalidRecordDoiExporterJob.xml
1
<RESOURCE_PROFILE>
2
    <HEADER>
3
        <RESOURCE_IDENTIFIER value="7edd375c-1d1e-4f2b-81c9-1b7d4f65c233_SGFkb29wSm9iQ29uZmlndXJhdGlvbkRTUmVzb3VyY2VzL0hhZG9vcEpvYkNvbmZpZ3VyYXRpb25EU1Jlc291cmNlVHlwZQ=="/>
4
        <RESOURCE_TYPE value="HadoopJobConfigurationDSResourceType"/>
5
        <RESOURCE_KIND value="HadoopJobConfigurationDSResources"/>
6
        <RESOURCE_URI value=""/>
7
        <DATE_OF_CREATION value="2018-10-22T10:34:48+00:00"/>
8
    </HEADER>
9
    <BODY>
10
        <HADOOP_JOB name="invalidRecordDoiExporterJob" type="mapreduce">
11
            <DESCRIPTION>map only job that ensures xml records are valid</DESCRIPTION>
12
            <STATIC_CONFIGURATION><!-- I/O FORMAT -->
13
                <PROPERTY key="mapreduce.inputformat.class" value="org.apache.hadoop.mapreduce.lib.input.SequenceFileInputFormat"/>
14
                <PROPERTY key="mapreduce.outputformat.class" value="org.apache.hadoop.mapreduce.lib.output.TextOutputFormat"/><!-- MAPPER -->
15
                <PROPERTY key="mapreduce.map.class" value="eu.dnetlib.data.mapreduce.hbase.dataimport.GetInvalidXmlRecordsMapper"/>
16
                <PROPERTY key="mapred.mapoutput.key.class" value="org.apache.hadoop.io.Text"/>
17
                <PROPERTY key="mapred.mapoutput.value.class" value="org.apache.hadoop.io.Text"/><!-- JOB GLOBAL -->
18
                <PROPERTY key="mapred.output.key.class" value="org.apache.hadoop.io.Text"/>
19
                <PROPERTY key="mapred.output.value.class" value="org.apache.hadoop.io.Text"/><!-- MISC -->
20
                <PROPERTY key="mapred.map.tasks.speculative.execution" value="false"/>
21
                <PROPERTY key="mapreduce.map.speculative" value="false"/>
22
                <PROPERTY key="mapred.reduce.tasks" value="1"/><!-- <PROPERTY key="user.name" value="dnet" /> --><!--  	Uncomment to override the default lib path -->
23
                <!-- <PROPERTY key="job.lib" value="/lib/dnet/snapshots/dnet-mapreduce-jobs-assembly-1.1.4-BETA-SNAPSHOT-FilterXmlRecordsMapper.jar"/> -->
24
                <PROPERTY key="mapred.output.compress" value="false"/>
25
            </STATIC_CONFIGURATION>
26
            <JOB_INTERFACE>
27
                <PARAM description="input sequence file" name="mapred.input.dir" required="true"/>
28
                <PARAM description="target file" name="mapred.output.dir" required="true"/>
29
            </JOB_INTERFACE>
30
        </HADOOP_JOB>
31
        <STATUS>
32
            <LAST_SUBMISSION_DATE value="2018-10-22T10:35:08+00:00"/>
33
            <RUNNING_INSTANCES value="0"/>
34
            <CUMULATIVE_RUN value="20"/>
35
        </STATUS>
36
        <SECURITY_PARAMETERS>SECURITY_PARAMETERS</SECURITY_PARAMETERS>
37
    </BODY>
38
</RESOURCE_PROFILE>
modules/dnet-openaireplus-profiles/trunk/src/main/resources/eu/dnetlib/test/profiles/HadoopJobConfigurationDSResources/HadoopJobConfigurationDSResourceType/prepareIndexDataJob.xml
32 32
				<PROPERTY key="mapred.map.tasks.speculative.execution" value="false" />	
33 33
				<PROPERTY key="mapreduce.map.speculative" value="false" />		
34 34
				<PROPERTY key="mapreduce.reduce.speculative" value="false" />
35
							
35

  
36
				<PROPERTY key="mapred.fairscheduler.pool" value="solr"/>
36 37
				<PROPERTY key="mapred.reduce.tasks" value="500" />
37 38
				<PROPERTY key="dfs.blocksize" value="256M" />
38 39
		        <PROPERTY key="ui.other.datasourcetypes" value="scholarcomminfra, infospace, pubsrepository::mock, entityregistry, entityregistry::projects, entityregistry::repositories, websource"/>
modules/dnet-openaireplus-profiles/trunk/src/main/resources/eu/dnetlib/test/profiles/HadoopJobConfigurationDSResources/HadoopJobConfigurationDSResourceType/xmlRecordValidatorJob.xml
1
<RESOURCE_PROFILE>
2
    <HEADER>
3
        <RESOURCE_IDENTIFIER value="d051952c-6f2f-4374-ac14-7fbb8d8fa430_SGFkb29wSm9iQ29uZmlndXJhdGlvbkRTUmVzb3VyY2VzL0hhZG9vcEpvYkNvbmZpZ3VyYXRpb25EU1Jlc291cmNlVHlwZQ=="/>
4
        <RESOURCE_TYPE value="HadoopJobConfigurationDSResourceType"/>
5
        <RESOURCE_KIND value="HadoopJobConfigurationDSResources"/>
6
        <RESOURCE_URI value=""/>
7
        <DATE_OF_CREATION value="2018-10-22T10:34:48+00:00"/>
8
    </HEADER>
9
    <BODY>
10
        <HADOOP_JOB name="xmlRecordValidatorJob" type="mapreduce">
11
            <DESCRIPTION>map only job that ensures xml records are valid</DESCRIPTION>
12
            <STATIC_CONFIGURATION><!-- I/O FORMAT -->
13
                <PROPERTY key="mapreduce.inputformat.class" value="org.apache.hadoop.mapreduce.lib.input.SequenceFileInputFormat"/>
14
                <PROPERTY key="mapreduce.outputformat.class" value="org.apache.hadoop.mapreduce.lib.output.SequenceFileOutputFormat"/><!-- MAPPER -->
15
                <PROPERTY key="mapreduce.map.class" value="eu.dnetlib.data.mapreduce.hbase.dataimport.FilterXmlRecordsMapper"/>
16
                <PROPERTY key="mapred.mapoutput.key.class" value="org.apache.hadoop.io.Text"/>
17
                <PROPERTY key="mapred.mapoutput.value.class" value="org.apache.hadoop.io.Text"/><!-- JOB GLOBAL -->
18
                <PROPERTY key="mapred.output.key.class" value="org.apache.hadoop.io.Text"/>
19
                <PROPERTY key="mapred.output.value.class" value="org.apache.hadoop.io.Text"/><!-- MISC -->
20
                <PROPERTY key="mapred.map.tasks.speculative.execution" value="false"/>
21
                <PROPERTY key="mapreduce.map.speculative" value="false"/>
22
                <PROPERTY key="mapred.reduce.tasks" value="100"/><!-- <PROPERTY key="user.name" value="dnet" /> --><!--  	Uncomment to override the default lib path -->
23
                <!-- <PROPERTY key="job.lib" value="/lib/dnet/snapshots/dnet-mapreduce-jobs-assembly-1.1.4-BETA-SNAPSHOT-FilterXmlRecordsMapper.jar"/> -->
24
                <PROPERTY key="mapred.output.compress" value="true"/>
25
                <PROPERTY key="mapred.output.compression.type" value="BLOCK"/>
26
            </STATIC_CONFIGURATION>
27
            <JOB_INTERFACE>
28
                <PARAM description="input sequence file" name="mapred.input.dir" required="true"/>
29
                <PARAM description="target file" name="mapred.output.dir" required="true"/>
30
            </JOB_INTERFACE>
31
        </HADOOP_JOB>
32
        <STATUS>
33
            <LAST_SUBMISSION_DATE value="2018-10-22T10:35:08+00:00"/>
34
            <RUNNING_INSTANCES value="0"/>
35
            <CUMULATIVE_RUN value="20"/>
36
        </STATUS>
37
        <SECURITY_PARAMETERS>SECURITY_PARAMETERS</SECURITY_PARAMETERS>
38
    </BODY>
39
</RESOURCE_PROFILE>
modules/dnet-openaireplus-profiles/trunk/src/main/resources/eu/dnetlib/test/profiles/HadoopJobConfigurationDSResources/HadoopJobConfigurationDSResourceType/informationSpaceConfigurableExportJob.xml
1
<RESOURCE_PROFILE>
2
    <HEADER>
3
        <RESOURCE_IDENTIFIER value="93078a14-40f4-49f0-9206-a16e6d4e342d_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="informationSpaceConfigurableExportJob" type="mapreduce">
11
 			<DESCRIPTION>map only job that exports the whole information space table as json</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.TextOutputFormat" />
17
        	
18
        		<!-- MAPPER -->
19
				<PROPERTY key="mapreduce.map.class" value="eu.dnetlib.data.mapreduce.hbase.dataexport.ConfigurableExportMapper" />
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
			
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

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

  
34
				<PROPERTY key="dfs.blocksize" value="256M" />
35
				
36
				<PROPERTY key="mapred.reduce.tasks" value="1" />
37
				<!-- <PROPERTY key="user.name" value="dnet" /> -->
38
				
39
		<!--  	Uncomment to override the default lib path -->			
40
		<!--	<PROPERTY key="job.lib" value="/user/dnet/dnet-mapreduce-jobs-0.0.2-SNAPSHOT-jar-with-dependencies.jar"/> -->
41
        	</STATIC_CONFIGURATION>
42
        	<JOB_INTERFACE>
43
        		<PARAM name="hbase.mapred.inputtable" required="true" description="source hbase table" />
44
        		<PARAM name="hbase.mapreduce.inputtable" required="true" description="source hbase table" />
45
        		
46
        		<PARAM name="mapred.output.dir" required="true" description="target sequence file on hdfs" />         		
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>

Also available in: Unified diff