628 lines
36 KiB
XML
628 lines
36 KiB
XML
<RESOURCE_PROFILE>
|
|
<HEADER>
|
|
<RESOURCE_IDENTIFIER value="4801c33c-66ca-4ab6-af64-aa812194ec67_V29ya2Zsb3dEU1Jlc291cmNlcy9Xb3JrZmxvd0RTUmVzb3VyY2VUeXBl"/>
|
|
<RESOURCE_TYPE value="WorkflowDSResourceType"/>
|
|
<RESOURCE_KIND value="WorkflowDSResources"/>
|
|
<RESOURCE_URI value=""/>
|
|
<DATE_OF_CREATION value="2021-07-30T09:42:23+00:00"/>
|
|
</HEADER>
|
|
<BODY>
|
|
<WORKFLOW_NAME>Graph construction for IIS [BETA]</WORKFLOW_NAME>
|
|
<WORKFLOW_TYPE>IIS</WORKFLOW_TYPE>
|
|
<WORKFLOW_PRIORITY>30</WORKFLOW_PRIORITY>
|
|
<CONFIGURATION start="manual">
|
|
<NODE isStart="true" name="setNsPrefixBlacklist" type="SetEnvParameter">
|
|
<DESCRIPTION>set blacklist of funder nsPrefixes</DESCRIPTION>
|
|
<PARAMETERS>
|
|
<PARAM managedBy="system" name="parameterName" required="true" type="string">nsPrefixBlacklist</PARAM>
|
|
<PARAM managedBy="user" name="parameterValue" required="true" type="string">gsrt________,rcuk________</PARAM>
|
|
</PARAMETERS>
|
|
<ARCS>
|
|
<ARC to="waitConfig"/>
|
|
</ARCS>
|
|
</NODE>
|
|
<NODE isStart="true" name="setIdMappingPath" type="SetEnvParameter">
|
|
<DESCRIPTION>set the path of the map defining the relations id mappings</DESCRIPTION>
|
|
<PARAMETERS>
|
|
<PARAM managedBy="system" name="parameterName" required="true" type="string">idMappingPath</PARAM>
|
|
<PARAM managedBy="user" name="parameterValue" required="true" type="string">/data/maps/fct_map.json</PARAM>
|
|
</PARAMETERS>
|
|
<ARCS>
|
|
<ARC to="waitConfig"/>
|
|
</ARCS>
|
|
</NODE>
|
|
<NODE isStart="true" name="setMergedGraphPath" type="SetEnvParameter">
|
|
<DESCRIPTION>Set the target path to store the MERGED graph</DESCRIPTION>
|
|
<PARAMETERS>
|
|
<PARAM managedBy="system" name="parameterName" required="true" type="string">mergedGraphPath</PARAM>
|
|
<PARAM managedBy="user" name="parameterValue" required="true" type="string">/tmp/beta_inference/graph/01_graph_merged</PARAM>
|
|
</PARAMETERS>
|
|
<ARCS>
|
|
<ARC to="waitConfig"/>
|
|
</ARCS>
|
|
</NODE>
|
|
<NODE isStart="true" name="setRawGraphPath" type="SetEnvParameter">
|
|
<DESCRIPTION>Set the target path to store the RAW graph</DESCRIPTION>
|
|
<PARAMETERS>
|
|
<PARAM managedBy="system" name="parameterName" required="true" type="string">rawGraphPath</PARAM>
|
|
<PARAM managedBy="user" name="parameterValue" required="true" type="string">/tmp/beta_inference/graph/02_graph_raw</PARAM>
|
|
</PARAMETERS>
|
|
<ARCS>
|
|
<ARC to="waitConfig"/>
|
|
</ARCS>
|
|
</NODE>
|
|
<NODE isStart="true" name="setCleanedFirstGraphPath" type="SetEnvParameter">
|
|
<DESCRIPTION>Set the target path to store the CLEANED graph</DESCRIPTION>
|
|
<PARAMETERS>
|
|
<PARAM managedBy="system" name="parameterName" required="true" type="string">cleanedFirstGraphPath</PARAM>
|
|
<PARAM managedBy="user" name="parameterValue" required="true" type="string">/tmp/beta_inference/graph/03_graph_clean_first</PARAM>
|
|
</PARAMETERS>
|
|
<ARCS>
|
|
<ARC to="waitConfig"/>
|
|
</ARCS>
|
|
</NODE>
|
|
<NODE isStart="true" name="setDedupGraphPath" type="SetEnvParameter">
|
|
<DESCRIPTION>Set the target path to store the DEDUPED graph</DESCRIPTION>
|
|
<PARAMETERS>
|
|
<PARAM managedBy="system" name="parameterName" required="true" type="string">dedupGraphPath</PARAM>
|
|
<PARAM managedBy="user" name="parameterValue" required="true" type="string">/tmp/beta_inference/graph/04_graph_dedup</PARAM>
|
|
</PARAMETERS>
|
|
<ARCS>
|
|
<ARC to="waitConfig"/>
|
|
</ARCS>
|
|
</NODE>
|
|
<NODE isStart="true" name="setConsistentGraphPath" type="SetEnvParameter">
|
|
<DESCRIPTION>Set the target path to store the CONSISTENCY graph</DESCRIPTION>
|
|
<PARAMETERS>
|
|
<PARAM managedBy="system" name="parameterName" required="true" type="string">consistentGraphPath</PARAM>
|
|
<PARAM managedBy="user" name="parameterValue" required="true" type="string">/tmp/beta_inference/graph/05_graph_consistent</PARAM>
|
|
</PARAMETERS>
|
|
<ARCS>
|
|
<ARC to="waitConfig"/>
|
|
</ARCS>
|
|
</NODE>
|
|
<NODE isStart="true" name="setCleanedGraphPath" type="SetEnvParameter">
|
|
<DESCRIPTION>Set the target path to store the CLEANED graph</DESCRIPTION>
|
|
<PARAMETERS>
|
|
<PARAM managedBy="system" name="parameterName" required="true" type="string">cleanedGraphPath</PARAM>
|
|
<PARAM managedBy="user" name="parameterValue" required="true" type="string">/tmp/beta_inference/graph/06_graph_cleaned</PARAM>
|
|
</PARAMETERS>
|
|
<ARCS>
|
|
<ARC to="waitConfig"/>
|
|
</ARCS>
|
|
</NODE>
|
|
<NODE isStart="true" name="setDedupConfig" type="SetEnvParameter">
|
|
<DESCRIPTION>Set the dedup orchestrator name</DESCRIPTION>
|
|
<PARAMETERS>
|
|
<PARAM managedBy="system" name="parameterName" required="true" type="string">dedupConfig</PARAM>
|
|
<PARAM managedBy="user" name="parameterValue" required="true" type="string">dedup-similarity-result-decisiontree-v2</PARAM>
|
|
</PARAMETERS>
|
|
<ARCS>
|
|
<ARC to="waitConfig"/>
|
|
</ARCS>
|
|
</NODE>
|
|
<NODE isStart="true" name="actionSetsRaw" type="SetEnvParameter">
|
|
<DESCRIPTION>declares the ActionSet ids to promote in the RAW graph</DESCRIPTION>
|
|
<PARAMETERS>
|
|
<PARAM managedBy="system" name="parameterName" required="true" type="string">actionSetIdsRawGraph</PARAM>
|
|
<PARAM managedBy="user" name="parameterValue" required="true" type="string">scholexplorer-dump,doiboost,orcidworks-no-doi,datacite</PARAM>
|
|
</PARAMETERS>
|
|
<ARCS>
|
|
<ARC to="waitConfig"/>
|
|
</ARCS>
|
|
</NODE>
|
|
<NODE isStart="true" name="setIsLookUpUrl" type="SetEnvParameter">
|
|
<DESCRIPTION>Set the IS lookup service address</DESCRIPTION>
|
|
<PARAMETERS>
|
|
<PARAM managedBy="system" name="parameterName" required="true" type="string">isLookUpUrl</PARAM>
|
|
<PARAM managedBy="system" name="parameterValue" required="true" type="string">http://beta.services.openaire.eu:8280/is/services/isLookUp?wsdl</PARAM>
|
|
</PARAMETERS>
|
|
<ARCS>
|
|
<ARC to="waitConfig"/>
|
|
</ARCS>
|
|
</NODE>
|
|
<NODE isJoin="true" name="waitConfig">
|
|
<DESCRIPTION>wait configurations</DESCRIPTION>
|
|
<PARAMETERS/>
|
|
<ARCS>
|
|
<ARC to="reuseODFClaims_PROD"/>
|
|
<ARC to="reuseODFClaims_BETA"/>
|
|
</ARCS>
|
|
</NODE>
|
|
<NODE name="reuseODFClaims_PROD" type="SetEnvParameter">
|
|
<DESCRIPTION>reuse cached ODF claims from the PROD aggregation system</DESCRIPTION>
|
|
<PARAMETERS>
|
|
<PARAM managedBy="system" name="parameterName" required="true" type="string">reuseODFClaims_PROD</PARAM>
|
|
<PARAM function="validValues(['true', 'false'])" managedBy="user" name="parameterValue" required="true" type="string">true</PARAM>
|
|
</PARAMETERS>
|
|
<ARCS>
|
|
<ARC to="reuseODF_hdfs_PROD"/>
|
|
</ARCS>
|
|
</NODE>
|
|
<NODE name="reuseODF_hdfs_PROD" type="SetEnvParameter">
|
|
<DESCRIPTION>reuse cached ODF records on HDFS from the PROD aggregation system</DESCRIPTION>
|
|
<PARAMETERS>
|
|
<PARAM managedBy="system" name="parameterName" required="true" type="string">reuseODFhdfs_PROD</PARAM>
|
|
<PARAM function="validValues(['true', 'false'])" managedBy="user" name="parameterValue" required="true" type="string">true</PARAM>
|
|
</PARAMETERS>
|
|
<ARCS>
|
|
<ARC to="reuseOAFClaims_PROD"/>
|
|
</ARCS>
|
|
</NODE>
|
|
<NODE name="reuseOAFClaims_PROD" type="SetEnvParameter">
|
|
<DESCRIPTION>reuse cached OAF claims from the PROD aggregation system</DESCRIPTION>
|
|
<PARAMETERS>
|
|
<PARAM managedBy="system" name="parameterName" required="true" type="string">reuseOAFClaims_PROD</PARAM>
|
|
<PARAM function="validValues(['true', 'false'])" managedBy="user" name="parameterValue" required="true" type="string">true</PARAM>
|
|
</PARAMETERS>
|
|
<ARCS>
|
|
<ARC to="reuseOAF_hdfs_PROD"/>
|
|
</ARCS>
|
|
</NODE>
|
|
<NODE name="reuseOAF_hdfs_PROD" type="SetEnvParameter">
|
|
<DESCRIPTION>reuse cached OAF records on HDFS from the PROD aggregation system</DESCRIPTION>
|
|
<PARAMETERS>
|
|
<PARAM managedBy="system" name="parameterName" required="true" type="string">reuseOAFhdfs_PROD</PARAM>
|
|
<PARAM function="validValues(['true', 'false'])" managedBy="user" name="parameterValue" required="true" type="string">true</PARAM>
|
|
</PARAMETERS>
|
|
<ARCS>
|
|
<ARC to="reuseDB_PROD"/>
|
|
</ARCS>
|
|
</NODE>
|
|
<NODE name="reuseDB_PROD" type="SetEnvParameter">
|
|
<DESCRIPTION>reuse cached DB content from the PROD aggregation system</DESCRIPTION>
|
|
<PARAMETERS>
|
|
<PARAM managedBy="system" name="parameterName" required="true" type="string">reuseDB_PROD</PARAM>
|
|
<PARAM function="validValues(['true', 'false'])" managedBy="user" name="parameterValue" required="true" type="string">true</PARAM>
|
|
</PARAMETERS>
|
|
<ARCS>
|
|
<ARC to="reuseDBOpenorgs_PROD"/>
|
|
</ARCS>
|
|
</NODE>
|
|
<NODE name="reuseDBOpenorgs_PROD" type="SetEnvParameter">
|
|
<DESCRIPTION>reuse cached OpenOrgs content from the PROD aggregation system</DESCRIPTION>
|
|
<PARAMETERS>
|
|
<PARAM managedBy="system" name="parameterName" required="true" type="string">reuseDBOpenorgs_PROD</PARAM>
|
|
<PARAM function="validValues(['true', 'false'])" managedBy="user" name="parameterValue" required="true" type="string">true</PARAM>
|
|
</PARAMETERS>
|
|
<ARCS>
|
|
<ARC to="reuseODF_PROD"/>
|
|
</ARCS>
|
|
</NODE>
|
|
<NODE name="reuseODF_PROD" type="SetEnvParameter">
|
|
<DESCRIPTION>reuse cached ODF content from the PROD aggregation system</DESCRIPTION>
|
|
<PARAMETERS>
|
|
<PARAM managedBy="system" name="parameterName" required="true" type="string">reuseODF_PROD</PARAM>
|
|
<PARAM function="validValues(['true', 'false'])" managedBy="user" name="parameterValue" required="true" type="string">true</PARAM>
|
|
</PARAMETERS>
|
|
<ARCS>
|
|
<ARC to="reuseOAF_PROD"/>
|
|
</ARCS>
|
|
</NODE>
|
|
<NODE name="reuseOAF_PROD" type="SetEnvParameter">
|
|
<DESCRIPTION>reuse cached OAF content from the PROD aggregation system</DESCRIPTION>
|
|
<PARAMETERS>
|
|
<PARAM managedBy="system" name="parameterName" required="true" type="string">reuseOAF_PROD</PARAM>
|
|
<PARAM function="validValues(['true', 'false'])" managedBy="user" name="parameterValue" required="true" type="string">true</PARAM>
|
|
</PARAMETERS>
|
|
<ARCS>
|
|
<ARC to="patchRelations_PROD"/>
|
|
</ARCS>
|
|
</NODE>
|
|
<NODE name="patchRelations_PROD" type="SetEnvParameter">
|
|
<DESCRIPTION>should apply the relations id patching based on the provided idMapping on PROD?</DESCRIPTION>
|
|
<PARAMETERS>
|
|
<PARAM managedBy="system" name="parameterName" required="true" type="string">shouldPatchRelations_PROD</PARAM>
|
|
<PARAM function="validValues(['true', 'false'])" managedBy="user" name="parameterValue" required="true" type="string">false</PARAM>
|
|
</PARAMETERS>
|
|
<ARCS>
|
|
<ARC to="contentPathProd"/>
|
|
</ARCS>
|
|
</NODE>
|
|
<NODE name="contentPathProd" type="SetEnvParameter">
|
|
<DESCRIPTION>set the PROD aggregator content path</DESCRIPTION>
|
|
<PARAMETERS>
|
|
<PARAM managedBy="system" name="parameterName" required="true" type="string">prodContentPath</PARAM>
|
|
<PARAM managedBy="user" name="parameterValue" required="true" type="string">/tmp/prod_aggregator_for_beta</PARAM>
|
|
</PARAMETERS>
|
|
<ARCS>
|
|
<ARC to="prodAggregatorGraphPath"/>
|
|
</ARCS>
|
|
</NODE>
|
|
<NODE name="prodAggregatorGraphPath" type="SetEnvParameter">
|
|
<DESCRIPTION>Set the path containing the PROD AGGREGATOR graph</DESCRIPTION>
|
|
<PARAMETERS>
|
|
<PARAM managedBy="system" name="parameterName" required="true" type="string">prodAggregatorGraphPath</PARAM>
|
|
<PARAM managedBy="user" name="parameterValue" required="true" type="string">/tmp/beta_inference/graph/00_prod_graph_aggregator</PARAM>
|
|
</PARAMETERS>
|
|
<ARCS>
|
|
<ARC to="waitConfig2"/>
|
|
</ARCS>
|
|
</NODE>
|
|
<NODE name="reuseODFClaims_BETA" type="SetEnvParameter">
|
|
<DESCRIPTION>reuse cached ODF claims from the BETA aggregation system</DESCRIPTION>
|
|
<PARAMETERS>
|
|
<PARAM managedBy="system" name="parameterName" required="true" type="string">reuseODFClaims_BETA</PARAM>
|
|
<PARAM function="validValues(['true', 'false'])" managedBy="user" name="parameterValue" required="true" type="string">true</PARAM>
|
|
</PARAMETERS>
|
|
<ARCS>
|
|
<ARC to="reuseODF_hdfs_BETA"/>
|
|
</ARCS>
|
|
</NODE>
|
|
<NODE name="reuseODF_hdfs_BETA" type="SetEnvParameter">
|
|
<DESCRIPTION>reuse cached ODF records on HDFS from the BETA aggregation system</DESCRIPTION>
|
|
<PARAMETERS>
|
|
<PARAM managedBy="system" name="parameterName" required="true" type="string">reuseODFhdfs_BETA</PARAM>
|
|
<PARAM function="validValues(['true', 'false'])" managedBy="user" name="parameterValue" required="true" type="string">true</PARAM>
|
|
</PARAMETERS>
|
|
<ARCS>
|
|
<ARC to="reuseOAFClaims_BETA"/>
|
|
</ARCS>
|
|
</NODE>
|
|
<NODE name="reuseOAFClaims_BETA" type="SetEnvParameter">
|
|
<DESCRIPTION>reuse cached OAF claims from the BETA aggregation system</DESCRIPTION>
|
|
<PARAMETERS>
|
|
<PARAM managedBy="system" name="parameterName" required="true" type="string">reuseOAFClaims_BETA</PARAM>
|
|
<PARAM function="validValues(['true', 'false'])" managedBy="user" name="parameterValue" required="true" type="string">true</PARAM>
|
|
</PARAMETERS>
|
|
<ARCS>
|
|
<ARC to="reuseOAF_hdfs_BETA"/>
|
|
</ARCS>
|
|
</NODE>
|
|
<NODE name="reuseOAF_hdfs_BETA" type="SetEnvParameter">
|
|
<DESCRIPTION>reuse cached OAF records on HDFS from the BETA aggregation system</DESCRIPTION>
|
|
<PARAMETERS>
|
|
<PARAM managedBy="system" name="parameterName" required="true" type="string">reuseOAFhdfs_BETA</PARAM>
|
|
<PARAM function="validValues(['true', 'false'])" managedBy="user" name="parameterValue" required="true" type="string">true</PARAM>
|
|
</PARAMETERS>
|
|
<ARCS>
|
|
<ARC to="reuseDB_BETA"/>
|
|
</ARCS>
|
|
</NODE>
|
|
<NODE name="reuseDB_BETA" type="SetEnvParameter">
|
|
<DESCRIPTION>reuse cached DB content from the BETA aggregation system</DESCRIPTION>
|
|
<PARAMETERS>
|
|
<PARAM managedBy="system" name="parameterName" required="true" type="string">reuseDB_BETA</PARAM>
|
|
<PARAM function="validValues(['true', 'false'])" managedBy="user" name="parameterValue" required="true" type="string">true</PARAM>
|
|
</PARAMETERS>
|
|
<ARCS>
|
|
<ARC to="reuseDBOpenorgs_BETA"/>
|
|
</ARCS>
|
|
</NODE>
|
|
<NODE name="reuseDBOpenorgs_BETA" type="SetEnvParameter">
|
|
<DESCRIPTION>reuse cached OpenOrgs content from the BETA aggregation system</DESCRIPTION>
|
|
<PARAMETERS>
|
|
<PARAM managedBy="system" name="parameterName" required="true" type="string">reuseDBOpenorgs_BETA</PARAM>
|
|
<PARAM function="validValues(['true', 'false'])" managedBy="user" name="parameterValue" required="true" type="string">true</PARAM>
|
|
</PARAMETERS>
|
|
<ARCS>
|
|
<ARC to="reuseODF_BETA"/>
|
|
</ARCS>
|
|
</NODE>
|
|
<NODE name="reuseODF_BETA" type="SetEnvParameter">
|
|
<DESCRIPTION>reuse cached ODF content from the BETA aggregation system</DESCRIPTION>
|
|
<PARAMETERS>
|
|
<PARAM managedBy="system" name="parameterName" required="true" type="string">reuseODF_BETA</PARAM>
|
|
<PARAM function="validValues(['true', 'false'])" managedBy="user" name="parameterValue" required="true" type="string">true</PARAM>
|
|
</PARAMETERS>
|
|
<ARCS>
|
|
<ARC to="reuseOAF_BETA"/>
|
|
</ARCS>
|
|
</NODE>
|
|
<NODE name="reuseOAF_BETA" type="SetEnvParameter">
|
|
<DESCRIPTION>reuse cached OAF content from the BETA aggregation system</DESCRIPTION>
|
|
<PARAMETERS>
|
|
<PARAM managedBy="system" name="parameterName" required="true" type="string">reuseOAF_BETA</PARAM>
|
|
<PARAM function="validValues(['true', 'false'])" managedBy="user" name="parameterValue" required="true" type="string">true</PARAM>
|
|
</PARAMETERS>
|
|
<ARCS>
|
|
<ARC to="patchRelations_BETA"/>
|
|
</ARCS>
|
|
</NODE>
|
|
<NODE name="patchRelations_BETA" type="SetEnvParameter">
|
|
<DESCRIPTION>should apply the relations id patching based on the provided idMapping on BETA?</DESCRIPTION>
|
|
<PARAMETERS>
|
|
<PARAM managedBy="system" name="parameterName" required="true" type="string">shouldPatchRelations_BETA</PARAM>
|
|
<PARAM function="validValues(['true', 'false'])" managedBy="user" name="parameterValue" required="true" type="string">false</PARAM>
|
|
</PARAMETERS>
|
|
<ARCS>
|
|
<ARC to="contentPathBeta"/>
|
|
</ARCS>
|
|
</NODE>
|
|
<NODE name="contentPathBeta" type="SetEnvParameter">
|
|
<DESCRIPTION>set the BETA aggregator content path</DESCRIPTION>
|
|
<PARAMETERS>
|
|
<PARAM managedBy="system" name="parameterName" required="true" type="string">betaContentPath</PARAM>
|
|
<PARAM managedBy="user" name="parameterValue" required="true" type="string">/tmp/beta_aggregator</PARAM>
|
|
</PARAMETERS>
|
|
<ARCS>
|
|
<ARC to="betaAggregatorGraphPath"/>
|
|
</ARCS>
|
|
</NODE>
|
|
<NODE name="betaAggregatorGraphPath" type="SetEnvParameter">
|
|
<DESCRIPTION>Set the path containing the BETA AGGREGATOR graph</DESCRIPTION>
|
|
<PARAMETERS>
|
|
<PARAM managedBy="system" name="parameterName" required="true" type="string">betaAggregatorGraphPath</PARAM>
|
|
<PARAM managedBy="user" name="parameterValue" required="true" type="string">/tmp/beta_inference/graph/00_beta_graph_aggregator</PARAM>
|
|
</PARAMETERS>
|
|
<ARCS>
|
|
<ARC to="waitConfig2"/>
|
|
</ARCS>
|
|
</NODE>
|
|
<NODE isJoin="true" name="waitConfig2">
|
|
<DESCRIPTION>wait configurations</DESCRIPTION>
|
|
<PARAMETERS/>
|
|
<ARCS>
|
|
<ARC to="betaAggregatorGraph"/>
|
|
<ARC to="prodAggregatorGraph"/>
|
|
</ARCS>
|
|
</NODE>
|
|
<NODE name="betaAggregatorGraph" type="SubmitHadoopJob">
|
|
<DESCRIPTION>create the BETA AGGREGATOR graph</DESCRIPTION>
|
|
<PARAMETERS>
|
|
<PARAM managedBy="system" name="hadoopJob" required="true" type="string">executeOozieJob</PARAM>
|
|
<PARAM managedBy="system" name="cluster" required="true" type="string">IIS</PARAM>
|
|
<PARAM managedBy="system" name="envParams" required="true" type="string">
|
|
{
|
|
'graphOutputPath' : 'betaAggregatorGraphPath',
|
|
'isLookupUrl' : 'isLookUpUrl',
|
|
'reuseODFClaims' : 'reuseODFClaims_BETA',
|
|
'reuseOAFClaims' : 'reuseOAFClaims_BETA',
|
|
'reuseDB' : 'reuseDB_BETA',
|
|
'reuseDBOpenorgs' : 'reuseDBOpenorgs_BETA',
|
|
'reuseODF' : 'reuseODF_BETA',
|
|
'reuseODF_hdfs' : 'reuseODFhdfs_BETA',
|
|
'reuseOAF' : 'reuseOAF_BETA',
|
|
'reuseOAF_hdfs' : 'reuseOAFhdfs_BETA',
|
|
'contentPath' : 'betaContentPath',
|
|
'nsPrefixBlacklist' : 'nsPrefixBlacklist',
|
|
'shouldPatchRelations' : 'shouldPatchRelations_BETA',
|
|
'idMappingPath' : 'idMappingPath'
|
|
}
|
|
</PARAM>
|
|
<PARAM managedBy="system" name="params" required="true" type="string">
|
|
{
|
|
'oozie.wf.application.path' : '/lib/dnet/BETA/oa/graph/raw_all/oozie_app',
|
|
'mongoURL' : '',
|
|
'mongoDb' : '',
|
|
'mdstoreManagerUrl' : '',
|
|
'postgresURL' : '',
|
|
'postgresUser' : '',
|
|
'postgresPassword' : '',
|
|
'postgresOpenOrgsURL' : '',
|
|
'postgresOpenOrgsUser' : '',
|
|
'postgresOpenOrgsPassword' : '',
|
|
'shouldHashId' : 'true',
|
|
'importOpenorgs' : 'true',
|
|
'workingDir' : '/tmp/beta_inference/working_dir/beta_aggregator'
|
|
}
|
|
</PARAM>
|
|
<PARAM managedBy="system" name="oozieReportActionsCsv" required="true" type="string">build-report</PARAM>
|
|
</PARAMETERS>
|
|
<ARCS>
|
|
<ARC to="waitAggregatorGraph"/>
|
|
</ARCS>
|
|
</NODE>
|
|
<NODE name="prodAggregatorGraph" type="SubmitHadoopJob">
|
|
<DESCRIPTION>create the PROD AGGREGATOR graph</DESCRIPTION>
|
|
<PARAMETERS>
|
|
<PARAM managedBy="system" name="hadoopJob" required="true" type="string">executeOozieJob</PARAM>
|
|
<PARAM managedBy="system" name="cluster" required="true" type="string">IIS</PARAM>
|
|
<PARAM managedBy="system" name="envParams" required="true" type="string">
|
|
{
|
|
'graphOutputPath' : 'prodAggregatorGraphPath',
|
|
'isLookupUrl' : 'isLookUpUrl',
|
|
'reuseODFClaims' : 'reuseODFClaims_PROD',
|
|
'reuseOAFClaims' : 'reuseOAFClaims_PROD',
|
|
'reuseDB' : 'reuseDB_PROD',
|
|
'reuseDBOpenorgs' : 'reuseDBOpenorgs_PROD',
|
|
'reuseODF' : 'reuseODF_PROD',
|
|
'reuseODF_hdfs' : 'reuseODFhdfs_PROD',
|
|
'reuseOAF' : 'reuseOAF_PROD',
|
|
'reuseOAF_hdfs' : 'reuseOAFhdfs_PROD',
|
|
'contentPath' : 'prodContentPath',
|
|
'nsPrefixBlacklist' : 'nsPrefixBlacklist',
|
|
'shouldPatchRelations' : 'shouldPatchRelations_PROD',
|
|
'idMappingPath' : 'idMappingPath'
|
|
}
|
|
</PARAM>
|
|
<PARAM managedBy="system" name="params" required="true" type="string">
|
|
{
|
|
'oozie.wf.application.path' : '/lib/dnet/BETA/oa/graph/raw_all/oozie_app',
|
|
'mongoURL' : '',
|
|
'mongoDb' : '',
|
|
'mdstoreManagerUrl' : '',
|
|
'postgresURL' : '',
|
|
'postgresUser' : '',
|
|
'postgresPassword' : '',
|
|
'postgresOpenOrgsURL' : '',
|
|
'postgresOpenOrgsUser' : '',
|
|
'postgresOpenOrgsPassword' : '',
|
|
'shouldHashId' : 'true',
|
|
'importOpenorgs' : 'true',
|
|
'workingDir' : '/tmp/beta_inference/working_dir/prod_aggregator'
|
|
}
|
|
</PARAM>
|
|
<PARAM managedBy="system" name="oozieReportActionsCsv" required="true" type="string">build-report</PARAM>
|
|
</PARAMETERS>
|
|
<ARCS>
|
|
<ARC to="waitAggregatorGraph"/>
|
|
</ARCS>
|
|
</NODE>
|
|
<NODE isJoin="true" name="waitAggregatorGraph">
|
|
<DESCRIPTION>wait configurations</DESCRIPTION>
|
|
<PARAMETERS/>
|
|
<ARCS>
|
|
<ARC to="mergeAggregatorGraphs"/>
|
|
</ARCS>
|
|
</NODE>
|
|
<NODE name="mergeAggregatorGraphs" type="SubmitHadoopJob">
|
|
<DESCRIPTION>create the AGGREGATOR graph</DESCRIPTION>
|
|
<PARAMETERS>
|
|
<PARAM managedBy="system" name="hadoopJob" required="true" type="string">executeOozieJob</PARAM>
|
|
<PARAM managedBy="system" name="cluster" required="true" type="string">IIS</PARAM>
|
|
<PARAM managedBy="system" name="envParams" required="true" type="string">
|
|
{
|
|
'betaInputGraphPath' : 'betaAggregatorGraphPath',
|
|
'prodInputGraphPath' : 'prodAggregatorGraphPath',
|
|
'graphOutputPath' : 'mergedGraphPath'
|
|
}
|
|
</PARAM>
|
|
<PARAM managedBy="system" name="params" required="true" type="string">
|
|
{
|
|
'oozie.wf.application.path' : '/lib/dnet/BETA/oa/graph/merge/oozie_app',
|
|
'workingDir' : '/tmp/beta_inference/working_dir/merge_graph',
|
|
'priority' : 'BETA'
|
|
}
|
|
</PARAM>
|
|
<PARAM managedBy="system" name="oozieReportActionsCsv" required="true" type="string">build-report</PARAM>
|
|
</PARAMETERS>
|
|
<ARCS>
|
|
<ARC to="promoteActionsRaw"/>
|
|
</ARCS>
|
|
</NODE>
|
|
<NODE name="promoteActionsRaw" type="SubmitHadoopJob">
|
|
<DESCRIPTION>create the RAW graph</DESCRIPTION>
|
|
<PARAMETERS>
|
|
<PARAM managedBy="system" name="hadoopJob" required="true" type="string">executeOozieJob</PARAM>
|
|
<PARAM managedBy="system" name="cluster" required="true" type="string">IIS</PARAM>
|
|
<PARAM managedBy="system" name="envParams" required="true" type="string">
|
|
{
|
|
'inputActionSetIds' : 'actionSetIdsRawGraph',
|
|
'inputGraphRootPath' : 'mergedGraphPath',
|
|
'outputGraphRootPath' : 'rawGraphPath',
|
|
'isLookupUrl' : 'isLookUpUrl'
|
|
}
|
|
</PARAM>
|
|
<PARAM managedBy="system" name="params" required="true" type="string">
|
|
{
|
|
'oozie.wf.application.path' : '/lib/dnet/BETA/actionmanager/wf/main/oozie_app',
|
|
'sparkExecutorCores' : '3',
|
|
'sparkExecutorMemory' : '10G',
|
|
'activePromoteDatasetActionPayload' : 'true',
|
|
'activePromoteDatasourceActionPayload' : 'true',
|
|
'activePromoteOrganizationActionPayload' : 'true',
|
|
'activePromoteOtherResearchProductActionPayload' : 'true',
|
|
'activePromoteProjectActionPayload' : 'true',
|
|
'activePromotePublicationActionPayload' : 'true',
|
|
'activePromoteRelationActionPayload' : 'true',
|
|
'activePromoteResultActionPayload' : 'true',
|
|
'activePromoteSoftwareActionPayload' : 'true',
|
|
'mergeAndGetStrategy' : 'MERGE_FROM_AND_GET',
|
|
'workingDir' : '/tmp/beta_inference/working_dir/promoteActionsRaw'
|
|
}
|
|
</PARAM>
|
|
<PARAM managedBy="system" name="oozieReportActionsCsv" required="true" type="string">build-report</PARAM>
|
|
</PARAMETERS>
|
|
<ARCS>
|
|
<ARC to="graphCleaningFirst"/>
|
|
</ARCS>
|
|
</NODE>
|
|
<NODE name="graphCleaningFirst" type="SubmitHadoopJob">
|
|
<DESCRIPTION>clean the properties in the graph typed as Qualifier according to the vocabulary indicated in schemeid</DESCRIPTION>
|
|
<PARAMETERS>
|
|
<PARAM managedBy="system" name="hadoopJob" required="true" type="string">executeOozieJob</PARAM>
|
|
<PARAM managedBy="system" name="cluster" required="true" type="string">IIS</PARAM>
|
|
<PARAM managedBy="system" name="envParams" required="true" type="string">
|
|
{
|
|
'graphInputPath' : 'rawGraphPath',
|
|
'graphOutputPath': 'cleanedFirstGraphPath',
|
|
'isLookupUrl': 'isLookUpUrl'
|
|
}
|
|
</PARAM>
|
|
<PARAM managedBy="system" name="params" required="true" type="string">
|
|
{
|
|
'oozie.wf.application.path' : '/lib/dnet/BETA/oa/graph/clean/oozie_app',
|
|
'workingDir' : '/tmp/beta_inference/working_dir/clean_first'
|
|
}
|
|
</PARAM>
|
|
<PARAM managedBy="system" name="oozieReportActionsCsv" required="true" type="string">build-report</PARAM>
|
|
</PARAMETERS>
|
|
<ARCS>
|
|
<ARC to="duplicateScan"/>
|
|
</ARCS>
|
|
</NODE>
|
|
<NODE name="duplicateScan" type="SubmitHadoopJob">
|
|
<DESCRIPTION>search for duplicates in the raw graph</DESCRIPTION>
|
|
<PARAMETERS>
|
|
<PARAM managedBy="system" name="hadoopJob" required="true" type="string">executeOozieJob</PARAM>
|
|
<PARAM managedBy="system" name="cluster" required="true" type="string">IIS</PARAM>
|
|
<PARAM managedBy="system" name="envParams" required="true" type="string">
|
|
{
|
|
'actionSetId' : 'dedupConfig',
|
|
'graphBasePath' : 'cleanedFirstGraphPath',
|
|
'dedupGraphPath': 'dedupGraphPath',
|
|
'isLookUpUrl' : 'isLookUpUrl'
|
|
}
|
|
</PARAM>
|
|
<PARAM managedBy="system" name="params" required="true" type="string">
|
|
{
|
|
'oozie.wf.application.path' : '/lib/dnet/BETA/oa/dedup/scan/oozie_app',
|
|
'actionSetIdOpenorgs' : 'dedup-similarity-organization-simple',
|
|
'workingPath' : '/tmp/beta_inference/working_dir/dedup',
|
|
'sparkExecutorCores' : '3',
|
|
'sparkExecutorMemory' : '10G'
|
|
}
|
|
</PARAM>
|
|
<PARAM managedBy="system" name="oozieReportActionsCsv" required="true" type="string">build-report</PARAM>
|
|
</PARAMETERS>
|
|
<ARCS>
|
|
<ARC to="dedupConsistency"/>
|
|
</ARCS>
|
|
</NODE>
|
|
<NODE name="dedupConsistency" type="SubmitHadoopJob">
|
|
<DESCRIPTION>mark duplicates as deleted and redistribute the relationships</DESCRIPTION>
|
|
<PARAMETERS>
|
|
<PARAM managedBy="system" name="hadoopJob" required="true" type="string">executeOozieJob</PARAM>
|
|
<PARAM managedBy="system" name="cluster" required="true" type="string">IIS</PARAM>
|
|
<PARAM managedBy="system" name="envParams" required="true" type="string">
|
|
{
|
|
'graphBasePath' : 'dedupGraphPath',
|
|
'graphOutputPath': 'consistentGraphPath'
|
|
}
|
|
</PARAM>
|
|
<PARAM managedBy="system" name="params" required="true" type="string">
|
|
{
|
|
'oozie.wf.application.path' : '/lib/dnet/BETA/oa/dedup/consistency/oozie_app',
|
|
'workingPath' : '/tmp/beta_inference/working_dir/dedup'
|
|
}
|
|
</PARAM>
|
|
<PARAM managedBy="system" name="oozieReportActionsCsv" required="true" type="string">build-report</PARAM>
|
|
</PARAMETERS>
|
|
<ARCS>
|
|
<ARC to="graphCleaning"/>
|
|
</ARCS>
|
|
</NODE>
|
|
<NODE name="graphCleaning" type="SubmitHadoopJob">
|
|
<DESCRIPTION>clean the properties in the graph typed as Qualifier according to the vocabulary indicated in schemeid</DESCRIPTION>
|
|
<PARAMETERS>
|
|
<PARAM managedBy="system" name="hadoopJob" required="true" type="string">executeOozieJob</PARAM>
|
|
<PARAM managedBy="system" name="cluster" required="true" type="string">IIS</PARAM>
|
|
<PARAM managedBy="system" name="envParams" required="true" type="string">
|
|
{
|
|
'graphInputPath' : 'consistentGraphPath',
|
|
'graphOutputPath': 'cleanedGraphPath',
|
|
'isLookupUrl': 'isLookUpUrl'
|
|
}
|
|
</PARAM>
|
|
<PARAM managedBy="system" name="params" required="true" type="string">
|
|
{
|
|
'oozie.wf.application.path' : '/lib/dnet/BETA/oa/graph/clean/oozie_app',
|
|
'workingDir' : '/tmp/beta_inference/working_dir/clean'
|
|
}
|
|
</PARAM>
|
|
<PARAM managedBy="system" name="oozieReportActionsCsv" required="true" type="string">build-report</PARAM>
|
|
</PARAMETERS>
|
|
<ARCS>
|
|
<ARC to="success"/>
|
|
</ARCS>
|
|
</NODE>
|
|
</CONFIGURATION>
|
|
<STATUS>
|
|
<LAST_EXECUTION_ID>wf_20210730_094240_462</LAST_EXECUTION_ID>
|
|
<LAST_EXECUTION_DATE>2021-07-30T15:04:19+00:00</LAST_EXECUTION_DATE>
|
|
<LAST_EXECUTION_STATUS>SUCCESS</LAST_EXECUTION_STATUS>
|
|
<LAST_EXECUTION_ERROR/>
|
|
</STATUS>
|
|
</BODY>
|
|
</RESOURCE_PROFILE> |