1
|
<RESOURCE_PROFILE>
|
2
|
<HEADER>
|
3
|
<RESOURCE_IDENTIFIER
|
4
|
value="f6e4cbcd-b173-4f8d-9205-e64ba15f03ad_SGFkb29wSm9iQ29uZmlndXJhdGlvbkRTUmVzb3VyY2VzL0hhZG9vcEpvYkNvbmZpZ3VyYXRpb25EU1Jlc291cmNlVHlwZQ=="/>
|
5
|
<RESOURCE_TYPE value="HadoopJobConfigurationDSResourceType"/>
|
6
|
<RESOURCE_KIND value="HadoopJobConfigurationDSResources"/>
|
7
|
<RESOURCE_URI value=""/>
|
8
|
<DATE_OF_CREATION value="2001-12-31T12:00:00"/>
|
9
|
</HEADER>
|
10
|
<BODY>
|
11
|
<HADOOP_JOB name="iisMainJob" type="oozie">
|
12
|
<DESCRIPTION>IIS main workflow</DESCRIPTION>
|
13
|
<STATIC_CONFIGURATION>
|
14
|
|
15
|
<!-- Cluster wide -->
|
16
|
<PROPERTY key="queueName" value="default"/>
|
17
|
<PROPERTY key="user.name" value="dnet.beta"/>
|
18
|
|
19
|
<!-- Runtime -->
|
20
|
<PROPERTY key="match_content_with_metadata" value="true"/>
|
21
|
<PROPERTY key="import_hbase_approved_datasources_csv" value="$UNDEFINED$"/>
|
22
|
<PROPERTY key="import_hbase_input_table" value="information_space-2014-11-05"/>
|
23
|
<!--<PROPERTY key="metadataextraction_default_cache_location" value="/cache/metadataextraction"/> -->
|
24
|
<PROPERTY key="metadataextraction_default_cache_location" value="/cache/metadataextraction_updated_model"/>
|
25
|
|
26
|
<PROPERTY key="export_action_hbase_table_initialize" value="true"/>
|
27
|
<PROPERTY key="import_content_connection_timeout" value="180000"/>
|
28
|
<PROPERTY key="import_content_read_timeout" value="180000"/>
|
29
|
<PROPERTY key="import_resultset_client_read_timeout" value="180000"/>
|
30
|
|
31
|
<PROPERTY key="oozie.wf.validate.ForkJoin" value="false"/>
|
32
|
<PROPERTY key="workingDir" value="/tmp/integration/working_dir/main"/>
|
33
|
<PROPERTY key="metadataextraction_excluded_checksums"
|
34
|
value="1e5b574109da731f4918c7f91fc24864,bea4728578070c3d66774bf9454d41fe,da458477233b5561ae47042aa2a73086"/>
|
35
|
<PROPERTY key="export_documentssimilarity_threshold" value="0.7"/>
|
36
|
<PROPERTY key="export_trust_level_threshold_document_classes" value="0.5"/>
|
37
|
<PROPERTY key="export_trust_level_threshold_document_referencedDatasets" value="0.0"/>
|
38
|
<PROPERTY key="export_trust_level_threshold_document_referencedProjects" value="0.0"/>
|
39
|
<PROPERTY key="export_trust_level_threshold_document_pdb" value="0.0"/>
|
40
|
<PROPERTY key="output_remote_location" value="hdfs://namenode.hadoop.dm.openaire.eu/var/lib/dnet/actionManager_BETA"/>
|
41
|
</STATIC_CONFIGURATION>
|
42
|
<JOB_INTERFACE>
|
43
|
<PARAM name="import_content_object_store_location" required="true" description="objectStore service endpoint"/>
|
44
|
<PARAM name="import_content_objectstores_csv" required="true" description="csv list of the available object stores subject to processing"/>
|
45
|
<PARAM name="import_mdstore_service_location" required="true" description="mdstore service location"/>
|
46
|
<PARAM name="import_dataset_mdstore_ids_csv" required="true" description="mdstore ids for dataset records"/>
|
47
|
<PARAM name="export_action_hbase_table_name" required="true" description="destination action manager table"/>
|
48
|
<PARAM name="export_action_hbase_remote_zookeeper_quorum" required="true" description="ZK quorum"/>
|
49
|
<PARAM name="export_action_hbase_remote_zookeeper_clientport" required="true" description="ZK port"/>
|
50
|
<PARAM name="nameNode" required="true" description="hdfs name node"/>
|
51
|
<PARAM name="jobTracker" required="true" description="job tracker name"/>
|
52
|
<PARAM name="oozie.wf.application.path" required="true" description="oozie job application absolute path"/>
|
53
|
<PARAM description="target action set for referenced projects" name="export_action_set_id_document_referencedProjects" required="true"/>
|
54
|
<PARAM description="target action set for referenced datasets" name="export_action_set_id_document_referencedDatasets" required="true"/>
|
55
|
<PARAM description="target action set for research initiative outcome" name="export_action_set_id_document_research_initiative"
|
56
|
required="true"/>
|
57
|
<PARAM description="target action set for document similarities" name="export_action_set_id_document_similarities_standard" required="true"/>
|
58
|
<PARAM description="target action set for document statistics" name="export_action_set_id_document_statistics" required="true"/>
|
59
|
<PARAM description="target action set for document classification" name="export_action_set_id_document_classes" required="true"/>
|
60
|
<PARAM description="target action set for document citations" name="export_action_set_id_document_referencedDocuments" required="true"/>
|
61
|
<PARAM description="target action set for dataset entities" name="export_action_set_id_entity_dataset" required="true"/>
|
62
|
<PARAM description="target action set for pdb entities" name="export_action_set_id_document_pdb" required="true"/>
|
63
|
|
64
|
<!-- flags to enable/disable IIS modules -->
|
65
|
<PARAM name="active_metadataextraction_export" required="true" description="enable/disable the affiliation matching module"/>
|
66
|
<PARAM name="active_citationmatching" required="true" description="enable/disable the citation matching module"/>
|
67
|
<PARAM name="active_documentsclassification" required="true" description="enable/disable the document classification module"/>
|
68
|
<PARAM name="active_documentssimilarity" required="true" description="enable/disable the document similarity module"/>
|
69
|
<PARAM name="active_referenceextraction_dataset" required="true" description="enable/disable the dataset reference extraction module"/>
|
70
|
<PARAM name="active_referenceextraction_project" required="true" description="enable/disable the project reference extracion module"/>
|
71
|
<PARAM name="active_referenceextraction_researchinitiative" required="true"
|
72
|
description="enable/disable the research initiative extraction module"/>
|
73
|
<PARAM name="active_statistics" required="true" description="enable/disable the statistics module"/>
|
74
|
<PARAM name="active_referenceextraction_pdb" required="true" description="enable/disable the protein data bank extraction module"/>
|
75
|
|
76
|
</JOB_INTERFACE>
|
77
|
</HADOOP_JOB>
|
78
|
<STATUS>
|
79
|
<LAST_SUBMISSION_DATE value="2001-12-31T12:00:00"/>
|
80
|
<RUNNING_INSTANCES value="0"/>
|
81
|
<CUMULATIVE_RUN value="0"/>
|
82
|
</STATUS>
|
83
|
<SECURITY_PARAMETERS>SECURITY_PARAMETERS</SECURITY_PARAMETERS>
|
84
|
</BODY>
|
85
|
</RESOURCE_PROFILE>
|