Project

General

Profile

« Previous | Next » 

Revision 48635

updated IIS main job configuration profile with the settings from production

View differences:

modules/dnet-openaireplus-profiles/trunk/src/main/resources/eu/dnetlib/test/profiles/HadoopJobConfigurationDSResources/HadoopJobConfigurationDSResourceType/iisMainJobV2.xml
1 1
<RESOURCE_PROFILE>
2 2
	<HEADER>
3
		<RESOURCE_IDENTIFIER
4
				value="d730a831-a982-4034-a890-de98fd972e87_SGFkb29wSm9iQ29uZmlndXJhdGlvbkRTUmVzb3VyY2VzL0hhZG9vcEpvYkNvbmZpZ3VyYXRpb25EU1Jlc291cmNlVHlwZQ=="/>
3
		<RESOURCE_IDENTIFIER value="d730a831-a982-4034-a890-de98fd972e87_SGFkb29wSm9iQ29uZmlndXJhdGlvbkRTUmVzb3VyY2VzL0hhZG9vcEpvYkNvbmZpZ3VyYXRpb25EU1Jlc291cmNlVHlwZQ=="/>
5 4
		<RESOURCE_TYPE value="HadoopJobConfigurationDSResourceType"/>
6 5
		<RESOURCE_KIND value="HadoopJobConfigurationDSResources"/>
7 6
		<RESOURCE_URI value=""/>
8
		<DATE_OF_CREATION value="2001-12-31T12:00:00"/>
7
		<DATE_OF_CREATION value="2017-07-05T15:05:50+00:00"/>
9 8
	</HEADER>
10 9
	<BODY>
11 10
		<HADOOP_JOB name="iisMainJobV2" type="oozie">
......
13 12
			<STATIC_CONFIGURATION>
14 13

  
15 14
				<!-- Cluster wide -->
16
				<PROPERTY key="user.name" value="dnet.beta"/>
15
				<PROPERTY key="user.name" value="dnet.production"/>
16

  
17 17
				<!-- Runtime -->
18 18
				<PROPERTY key="metadataextraction_default_cache_location" value="/user/${user.name}/iis/cache/metadataextraction"/>
19
				<PROPERTY key="ingest_pmc_default_cache_location" value="/user/${user.name}/iis/cache/ingestpmc"/>
19
				<!-- skipping _default chunk from property name since 2017.02.21 -->
20
				<PROPERTY key="metadataextraction_cache_location" value="/user/${user.name}/iis/cache/metadataextraction"/>
21
				<PROPERTY key="ingest_pmc_cache_location" value="/user/${user.name}/iis/cache/ingestpmc"/>
20 22
				<PROPERTY key="export_documentssimilarity_threshold" value="0.7"/>
21 23
				<PROPERTY key="export_trust_level_threshold_document_classes" value="0.5"/>
22 24
				<PROPERTY key="export_trust_level_threshold_document_referencedDatasets" value="0.0"/>
......
24 26
				<PROPERTY key="export_trust_level_threshold_document_pdb" value="0.0"/>
25 27
				<PROPERTY key="export_trust_level_threshold_document_software_url" value="0.0"/>
26 28
				<PROPERTY key="export_trust_level_threshold_matched_doc_organizations" value="0.0"/>
27
				<PROPERTY key="output_remote_location" value="webhdfs://namenode.hadoop.dm.openaire.eu:50071/var/lib/dnet/actionManager_BETA"/>
29
				<PROPERTY key="output_remote_location" value="webhdfs://namenode.hadoop.dm.openaire.eu:50071/var/lib/dnet/actionManager_PROD"/>
28 30
				<PROPERTY key="reports_external_path" value="/user/${user.name}/iis/reports/${execution_environment}"/>
29

  
30
				<PROPERTY key="import_hbase_dump_location" value="webhdfs://namenode.hadoop.dm.openaire.eu:50071/tmp/db_openaireplus_services.export.2016.09.15"/>
31
				<PROPERTY key="import_hbase_dump_location" value="webhdfs://namenode.hadoop.dm.openaire.eu:50071/tmp/db_openaireplus_services.export.2017.07.19"/>
31 32
				<PROPERTY key="oozieServiceLoc" value="http://iis-cdh5-test-m3:11000/oozie"/>
32 33
			</STATIC_CONFIGURATION>
33 34
			<JOB_INTERFACE>
35
				<PARAM description="oozie job application absolute path" name="oozie.wf.application.path" required="true"/>
36
				<PARAM description="objectStore service endpoint" name="import_content_object_store_location" required="true"/>
37
				<PARAM description="csv list of the available object stores subject to processing" name="import_content_objectstores_csv" required="true"/>
38
				<PARAM description="mdstore service location" name="import_mdstore_service_location" required="true"/>
39
				<PARAM description="mdstore ids for dataset records" name="import_dataset_mdstore_ids_csv" required="true"/>
40
				<PARAM description="target action set for referenced datasets" name="export_action_set_id_document_referencedDatasets" required="true"/>
41
				<PARAM description="target action set for referenced projects" name="export_action_set_id_document_referencedProjects" required="true"/>
42
				<PARAM description="target action set for research initiative outcome" name="export_action_set_id_document_research_initiative" required="true"/>
43
				<PARAM description="target action set for document similarities" name="export_action_set_id_document_similarities_standard" required="true"/>
44
				<PARAM description="target action set for document classification" name="export_action_set_id_document_classes" required="true"/>
45
				<PARAM description="target action set for document citations" name="export_action_set_id_document_referencedDocuments" required="true"/>
46
				<PARAM description="target action set for dataset entities" name="export_action_set_id_entity_dataset" required="true"/>
47
				<PARAM description="target action set for pdb entities" name="export_action_set_id_document_pdb" required="true"/>
48
				<PARAM description="software reference extraction actionset identifier" name="export_action_set_id_document_software_url" required="true"/>
49
				<PARAM description="target action set for affiliations" name="export_action_set_id_matched_doc_organizations" required="true"/>
34 50

  
35
				<PARAM name="import_hbase_dump_location" required="true" description="InfoSpace HBase JSON records dump (not default HBase export outcome!) remote location. Should include protocol prefix and namenode when copying from remote cluster."/>
36
				<PARAM name="oozie.wf.application.path"  required="true" description="oozie job application absolute path"/>
37 51

  
38
				<PARAM name="import_content_object_store_location" required="true" description="objectStore service endpoint"/>
39
				<PARAM name="import_content_objectstores_csv"      required="true" description="csv list of the available object stores subject to processing"/>
40
				<PARAM name="import_mdstore_service_location"      required="true" description="mdstore service location"/>
41
				<PARAM name="import_dataset_mdstore_ids_csv"       required="true" description="mdstore ids for dataset records"/>
42

  
43

  
44
				<PARAM name="export_action_set_id_document_referencedDatasets"  required="true" description="target action set for referenced datasets"/>
45
				<PARAM name="export_action_set_id_document_referencedProjects"  required="true" description="target action set for referenced projects"/>
46
				<PARAM name="export_action_set_id_document_research_initiative" required="true" description="target action set for research initiative outcome"/>
47
				<PARAM name="export_action_set_id_document_similarities_standard" required="true" description="target action set for document similarities"   />
48
				<PARAM name="export_action_set_id_document_classes"               required="true" description="target action set for document classification" />
49
				<PARAM name="export_action_set_id_document_referencedDocuments"   required="true" description="target action set for document citations"      />
50
				<PARAM name="export_action_set_id_entity_dataset"                 required="true" description="target action set for dataset entities"        />
51
				<PARAM name="export_action_set_id_document_pdb"                   required="true" description="target action set for pdb entities"            />
52
				<PARAM name="export_action_set_id_document_software_url"          required="true" description="software reference extraction actionset identifier"/>
53
				<PARAM name="export_action_set_id_matched_doc_organizations"      required="true" description="target action set for affiliations"/>
54

  
55

  
56 52
				<!-- flags to enable/disable IIS modules -->
57
				<PARAM name="active_citationmatching"                       required="true" description="enable/disable the citation matching module"/>
58
				<PARAM name="active_documentsclassification"                required="true" description="enable/disable the document classification module"/>
59
				<PARAM name="active_documentssimilarity"                    required="true" description="enable/disable the document similarity module"/>
60
				<PARAM name="active_referenceextraction_dataset"            required="true" description="enable/disable the dataset reference extraction module"/>
61
				<PARAM name="active_referenceextraction_project"            required="true" description="enable/disable the project reference extracion module"/>
62
				<PARAM name="active_referenceextraction_researchinitiative" required="true" description="enable/disable the research initiative extraction module"/>
63
				<PARAM name="active_referenceextraction_pdb"                required="true" description="enable/disable the protein data bank extraction module"/>
64
				<PARAM name="active_referenceextraction_software_url"       required="true" description="enable/disable the software reference extraction algorithm"/>
65

  
53
				<PARAM description="enable/disable the citation matching module" name="active_citationmatching" required="true"/>
54
				<PARAM description="enable/disable the document classification module" name="active_documentsclassification" required="true"/>
55
				<PARAM description="enable/disable the document similarity module" name="active_documentssimilarity" required="true"/>
56
				<PARAM description="enable/disable the dataset reference extraction module" name="active_referenceextraction_dataset" required="true"/>
57
				<PARAM description="enable/disable the project reference extracion module" name="active_referenceextraction_project" required="true"/>
58
				<PARAM description="enable/disable the research initiative extraction module" name="active_referenceextraction_researchinitiative" required="true"/>
59
				<PARAM description="enable/disable the protein data bank extraction module" name="active_referenceextraction_pdb" required="true"/>
60
				<PARAM description="enable/disable the software reference extraction algorithm" name="active_referenceextraction_software_url" required="true"/>
66 61
			</JOB_INTERFACE>
67 62
		</HADOOP_JOB>
68 63
		<STATUS>
69
			<LAST_SUBMISSION_DATE value="2001-12-31T12:00:00"/>
64
			<LAST_SUBMISSION_DATE value="2017-07-02T09:39:51+00:00"/>
70 65
			<RUNNING_INSTANCES value="0"/>
71
			<CUMULATIVE_RUN value="0"/>
66
			<CUMULATIVE_RUN value="13"/>
72 67
		</STATUS>
73 68
		<SECURITY_PARAMETERS>SECURITY_PARAMETERS</SECURITY_PARAMETERS>
74 69
	</BODY>
75 70
</RESOURCE_PROFILE>
71
		production

Also available in: Unified diff