Project

General

Profile

« Previous | Next » 

Revision 56002

added workflow configuration and hadoop job for orcid publications without doi import

View differences:

modules/dnet-openaireplus-workflows/trunk/src/main/resources/eu/dnetlib/test/profiles/openaireplus/workflows/dm/import.infospace.xml
12 12
		<METAWORKFLOW_SECTION>InfoSpace Provision</METAWORKFLOW_SECTION>
13 13
		<ADMIN_EMAIL>alessia.bardi@isti.cnr.it,claudio.atzori@isti.cnr.it</ADMIN_EMAIL>
14 14
		<CONFIGURATION status="EXECUTABLE">
15
			<WORKFLOW id="e03f256e-1e4d-4b3d-9c07-91faf5d25207_V29ya2Zsb3dEU1Jlc291cmNlcy9Xb3JrZmxvd0RTUmVzb3VyY2VUeXBl" name="Import ScholExplorer">
16
			</WORKFLOW>
17
		</CONFIGURATION>
15
            <WORKFLOW id="e03f256e-1e4d-4b3d-9c07-91faf5d25207_V29ya2Zsb3dEU1Jlc291cmNlcy9Xb3JrZmxvd0RTUmVzb3VyY2VUeXBl" name="Import ScholExplorer">
16
                <WORKFLOW id="e03f256e-1e4d-4b3d-9c07-91faf5d25208_V29ya2Zsb3dEU1Jlc291cmNlcy9Xb3JrZmxvd0RTUmVzb3VyY2VUeXBl" name="Import DOIboost"/>
17
                <WORKFLOW id="970c4826-cf6d-4dfe-850c-41e508d341fa_V29ya2Zsb3dEU1Jlc291cmNlcy9Xb3JrZmxvd0RTUmVzb3VyY2VUeXBl" name="Import organizations from DOIboost"/>
18
                <WORKFLOW id="ff25219c-e485-4440-8bc0-a8bbe42512ac_V29ya2Zsb3dEU1Jlc291cmNlcy9Xb3JrZmxvd0RTUmVzb3VyY2VUeXBl" name="Import Grid.AC"/>
19
                <WORKFLOW id="7c8765af-1253-4bd7-8806-315b73bf7319_V29ya2Zsb3dEU1Jlc291cmNlcy9Xb3JrZmxvd0RTUmVzb3VyY2VUeXBl" name="Import Orcid"/>
20
            </WORKFLOW>
21
        </CONFIGURATION>
18 22
		<SCHEDULING enabled="false">
19 23
			<CRON>29 5 22 ? * *</CRON>
20 24
			<MININTERVAL>10080</MININTERVAL>
modules/dnet-openaireplus-workflows/trunk/src/main/resources/eu/dnetlib/test/profiles/openaireplus/workflows/hbase/importOrcid.xml
1
<RESOURCE_PROFILE>
2
	<HEADER>
3
		<RESOURCE_IDENTIFIER value="7c8765af-1253-4bd7-8806-315b73bf7319_V29ya2Zsb3dEU1Jlc291cmNlcy9Xb3JrZmxvd0RTUmVzb3VyY2VUeXBl"/>
4
		<RESOURCE_TYPE value="WorkflowDSResourceType"/>
5
		<RESOURCE_KIND value="WorkflowDSResources"/>
6
		<RESOURCE_URI value=""/>
7
		<DATE_OF_CREATION value="2019-05-29T10:54:33+00:00"/>
8
	</HEADER>
9
	<BODY>
10
		<WORKFLOW_NAME>Import Orcid</WORKFLOW_NAME>
11
		<WORKFLOW_TYPE>Import InfoSpace</WORKFLOW_TYPE>
12
		<WORKFLOW_PRIORITY>30</WORKFLOW_PRIORITY>
13
		<CONFIGURATION start="manual">
14
			<NODE isStart="true" name="setInputPath" type="SetHdfsFile">
15
				<DESCRIPTION>set the hdfs output path</DESCRIPTION>
16
				<PARAMETERS>
17
					<PARAM managedBy="user" name="hdfsPath" required="true" type="string">/usr/dnet/data/orcid</PARAM>
18
					<PARAM managedBy="system" name="hdfsPathParam" required="true" type="string">inputPath</PARAM>
19
				</PARAMETERS>
20
				<ARCS>
21
					<ARC to="prepareActionSets"/>
22
				</ARCS>
23
			</NODE>
24
			<NODE name="prepareActionSets" type="PrepareActionSets">
25
				<DESCRIPTION>prepare action sets</DESCRIPTION>
26
				<PARAMETERS>
27
					<PARAM managedBy="system" name="sets" required="true" type="string">
28
						[
29
						{
30
						'set' : 'orcidworks-no-doi',
31
						'jobProperty' : 'export_action_set_orcidworks_no_doi',
32
						'enablingProperty' : 'active_orcidworks_no_doi',
33
						'enabled' : 'true'
34
						}
35
						]
36
					</PARAM>
37
				</PARAMETERS>
38
				<ARCS>
39
					<ARC to="extractOutputPath"/>
40
				</ARCS>
41
			</NODE>
42
			<NODE name="extractOutputPath" type="ExtractOutputPath">
43
				<DESCRIPTION>extract the hdfs output path generated in the previous node</DESCRIPTION>
44
				<PARAMETERS>
45
					<PARAM managedBy="system" name="hdfsOutputPathParam" required="true" type="string">outputPath</PARAM>
46
				</PARAMETERS>
47
				<ARCS>
48
					<ARC to="importActionSet"/>
49
				</ARCS>
50
			</NODE>
51
			<NODE isJoin="true" name="importActionSet" type="SubmitHadoopJob">
52
				<DESCRIPTION>IIS main</DESCRIPTION>
53
				<PARAMETERS>
54
					<PARAM managedBy="system" name="hadoopJob" required="true" type="string">importOrcidJob</PARAM>
55
					<PARAM managedBy="system" name="cluster" required="true" type="string">DM</PARAM>
56
					<PARAM managedBy="system" name="envParams" required="true" type="string">
57
						{
58
						'mapred.input.dir':'inputPath',
59
						'mapred.output.dir':'outputPath'
60
						}
61
					</PARAM>
62
					<PARAM managedBy="user" name="simulation" required="true" type="boolean">false</PARAM>
63
				</PARAMETERS>
64
				<ARCS>
65
					<ARC to="updateActionSets"/>
66
				</ARCS>
67
			</NODE>
68
			<NODE name="updateActionSets" type="UpdateActionSets">
69
				<DESCRIPTION>update action sets</DESCRIPTION>
70
				<PARAMETERS/>
71
				<ARCS>
72
					<ARC to="success"/>
73
				</ARCS>
74
			</NODE>
75
		</CONFIGURATION>
76
		<STATUS/>
77
	</BODY>
78
</RESOURCE_PROFILE>
modules/dnet-openaireplus-profiles/trunk/src/main/resources/eu/dnetlib/test/profiles/HadoopJobConfigurationDSResources/HadoopJobConfigurationDSResourceType/importOrcidJob.xml
1
<RESOURCE_PROFILE>
2
	<HEADER>
3
		<RESOURCE_IDENTIFIER
4
				value="249e40c2-6420-4207-b40a-e1236f77f1fc_SGFkb29wSm9iQ29uZmlndXJhdGlvbkRTUmVzb3VyY2VzL0hhZG9vcEpvYkNvbmZpZ3VyYXRpb25EU1Jlc291cmNlVHlwZQ=="/>
5
		<RESOURCE_TYPE value="HadoopJobConfigurationDSResourceType"/>
6
		<RESOURCE_KIND value="HadoopJobConfigurationDSResources"/>
7
		<RESOURCE_URI value=""/>
8
		<DATE_OF_CREATION value="2019-05-29T12:00:00"/>
9
	</HEADER>
10
	<BODY>
11
		<HADOOP_JOB name="importOrcidJob" type="mapreduce">
12
			<DESCRIPTION>map reduce job that import the Orcid works (no doi) into actions</DESCRIPTION>
13
			<STATIC_CONFIGURATION>
14

  
15
				<!-- I/O FORMAT -->
16
				<PROPERTY key="mapreduce.inputformat.class" value="org.apache.hadoop.mapreduce.lib.input.TextInputFormat"/>
17
				<PROPERTY key="mapreduce.outputformat.class" value="org.apache.hadoop.mapreduce.lib.output.SequenceFileOutputFormat"/>
18

  
19
				<!-- MAPPER -->
20
				<PROPERTY key="mapreduce.map.class" value="eu.dnetlib.data.mapreduce.hbase.dataimport.OrcidImportMapper"/>
21
				<PROPERTY key="mapred.mapoutput.key.class" value="org.apache.hadoop.io.LongWritable"/>
22
				<PROPERTY key="mapred.mapoutput.value.class" value="org.apache.hadoop.io.Text"/>
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
				<PROPERTY key="mapred.compress.map.output" value="false"/>
29
				<PROPERTY key="mapred.map.tasks.speculative.execution" value="false"/>
30
				<PROPERTY key="mapreduce.map.speculative" value="false"/>
31

  
32
				<PROPERTY key="mapred.reduce.tasks" value="0"/>
33
				<PROPERTY key="dfs.blocksize" value="256M"/>
34

  
35
                <!-- Orcid Mapper Properties -->
36
                <PROPERTY key="setName" value="orcidworks-no-doi"/>
37
				<PROPERTY key="agentId" value="dnet"/>
38
				<PROPERTY key="agentName" value="D-Net"/>
39
				<PROPERTY key="invisible" value="true"/>
40

  
41
				<!-- <PROPERTY key="user.name" value="dnet" /> -->
42

  
43
				<!--  	Uncomment to override the default lib path -->
44
				<!--  PROPERTY key="job.lib" value="/lib/dnet/snapshots/dnet-mapreduce-jobs-orcid_beta.jar"/ -->
45
			</STATIC_CONFIGURATION>
46
			<JOB_INTERFACE>
47
				<PARAM name="mapred.input.dir" required="true" description="source sequence file on hdfs"/>
48
				<PARAM name="mapred.output.dir" required="true" description="target sequence file on hdfs"/>
49
			</JOB_INTERFACE>
50
			<SCAN>
51
				<FILTERS/>
52
				<FAMILIES/>
53
			</SCAN>
54
		</HADOOP_JOB>
55
		<STATUS/>
56
		<SECURITY_PARAMETERS>SECURITY_PARAMETERS</SECURITY_PARAMETERS>
57
	</BODY>
58
</RESOURCE_PROFILE>
59

  

Also available in: Unified diff