Revision 58329
Added by Claudio Atzori about 4 years ago
modules/dnet-openaireplus-workflows/trunk/src/main/resources/eu/dnetlib/test/profiles/openaireplus/workflows/dedup/dedup.ocean.xml | ||
---|---|---|
1 |
<RESOURCE_PROFILE> |
|
2 |
<HEADER> |
|
3 |
<RESOURCE_IDENTIFIER value="2483c5e4-dde8-4452-bd9c-38d436d8d941_V29ya2Zsb3dEU1Jlc291cmNlcy9Xb3JrZmxvd0RTUmVzb3VyY2VUeXBl"/> |
|
4 |
<RESOURCE_TYPE value="WorkflowDSResourceType"/> |
|
5 |
<RESOURCE_KIND value="WorkflowDSResources"/> |
|
6 |
<RESOURCE_URI value=""/> |
|
7 |
<DATE_OF_CREATION value="2020-03-20T22:49:01+00:00"/> |
|
8 |
</HEADER> |
|
9 |
<BODY> |
|
10 |
<WORKFLOW_NAME>Offline Deduplication [OCEAN]</WORKFLOW_NAME> |
|
11 |
<WORKFLOW_TYPE>Deduplication</WORKFLOW_TYPE> |
|
12 |
<WORKFLOW_PRIORITY>30</WORKFLOW_PRIORITY> |
|
13 |
<CONFIGURATION start="manual"> |
|
14 |
|
|
15 |
<NODE name="setGraphBasePath" type="SetEnvParameter" isStart="true"> |
|
16 |
<DESCRIPTION>Set the target directory to materialize the graph</DESCRIPTION> |
|
17 |
<PARAMETERS> |
|
18 |
<PARAM managedBy="system" name="parameterName" required="true" type="string">graphBasePath</PARAM> |
|
19 |
<PARAM managedBy="user" name="parameterValue" required="true" type="string">/var/lib/dnet/graph_BETA</PARAM> |
|
20 |
</PARAMETERS> |
|
21 |
<ARCS> |
|
22 |
<ARC to="REUSE_RAW_GRAPH"/> |
|
23 |
</ARCS> |
|
24 |
</NODE> |
|
25 |
|
|
26 |
<NODE name="REUSE_RAW_GRAPH" type="Selection"> |
|
27 |
<DESCRIPTION>Is the raw graph already available?</DESCRIPTION> |
|
28 |
<PARAMETERS> |
|
29 |
<PARAM function="validValues(['YES', 'NO'])" managedBy="user" name="selection" required="true" type="string">YES</PARAM> |
|
30 |
</PARAMETERS> |
|
31 |
<ARCS> |
|
32 |
<ARC name="YES" to="setActionSetId"/> |
|
33 |
<ARC name="NO" to="rawGraph"/> |
|
34 |
</ARCS> |
|
35 |
</NODE> |
|
36 |
|
|
37 |
<NODE name="rawGraph" type="SubmitHadoopJob"> |
|
38 |
<DESCRIPTION>create the raw graph</DESCRIPTION> |
|
39 |
<PARAMETERS> |
|
40 |
<PARAM required="true" type="string" name="hadoopJob" managedBy="system">executeOozieJob</PARAM> |
|
41 |
<PARAM required="true" type="string" name="cluster" managedBy="system">IIS</PARAM> |
|
42 |
<PARAM required="true" type="string" name="envParams" managedBy="system"> |
|
43 |
{ |
|
44 |
'graphBasePath' : 'graphBasePath' |
|
45 |
} |
|
46 |
</PARAM> |
|
47 |
<PARAM required="true" type="string" name="params" managedBy="system"> |
|
48 |
{ |
|
49 |
'oozie.wf.application.path' : '/lib/dnet/migration/wfs/regular_all_steps/oozie_app', |
|
50 |
'mongoURL' : 'mongodb://beta.services.openaire.eu', |
|
51 |
'mongoDb' : 'mdstore', |
|
52 |
'postgresURL' : 'jdbc:postgresql://beta.services.openaire.eu:5432/dnet_openaireplus', |
|
53 |
'postgresUser' : 'dnet', |
|
54 |
'postgresPassword' : 'xxxx', |
|
55 |
'workingPath' : '/tmp/dhp_migration_beta', |
|
56 |
'reuseContent' : 'true' |
|
57 |
} |
|
58 |
</PARAM> |
|
59 |
<PARAM required="true" type="boolean" name="simulation" managedBy="system">false</PARAM> |
|
60 |
<PARAM required="true" type="string" name="oozieReportActionsCsv" managedBy="system">build-report</PARAM> |
|
61 |
</PARAMETERS> |
|
62 |
<ARCS> |
|
63 |
<ARC to="setActionSetId"/> |
|
64 |
</ARCS> |
|
65 |
</NODE> |
|
66 |
|
|
67 |
<NODE name="setActionSetId" type="SetEnvParameter"> |
|
68 |
<DESCRIPTION>Set action set id</DESCRIPTION> |
|
69 |
<PARAMETERS> |
|
70 |
<PARAM managedBy="system" name="parameterName" required="true" type="string">setActionSetId</PARAM> |
|
71 |
<PARAM function="obtainValues('dedupOrchestrations', {})" managedBy="user" name="parameterValue" required="true" type="string">dedup-similarity-organization</PARAM> |
|
72 |
</PARAMETERS> |
|
73 |
<ARCS> |
|
74 |
<ARC to="setDedupConfig"/> |
|
75 |
</ARCS> |
|
76 |
</NODE> |
|
77 |
|
|
78 |
<NODE name="setDedupConfig" type="SetDedupConfiguration"> |
|
79 |
<DESCRIPTION>Set Dedup conf</DESCRIPTION> |
|
80 |
<PARAMETERS> |
|
81 |
<PARAM function="obtainValues('dedupOrchestrations', {})" managedBy="user" name="dedupConfigSequence" required="true" type="string">dedup-similarity-organization</PARAM> |
|
82 |
<PARAM managedBy="system" name="dedupConfigSequenceParam" required="true" type="string">dedup.conf.queue</PARAM> |
|
83 |
</PARAMETERS> |
|
84 |
<ARCS> |
|
85 |
<ARC to="prepareActionSet"/> |
|
86 |
</ARCS> |
|
87 |
</NODE> |
|
88 |
|
|
89 |
<NODE name="prepareActionSet" type="PrepareConfiguredActionSet"> |
|
90 |
<DESCRIPTION>prepare action sets</DESCRIPTION> |
|
91 |
<PARAMETERS> |
|
92 |
<PARAM managedBy="system" name="dedupConfigSequenceParam" required="true" type="string">dedup.conf.queue</PARAM> |
|
93 |
<PARAM managedBy="system" name="jobProperty" required="true" type="string">rawSetId</PARAM> |
|
94 |
<PARAM managedBy="system" name="actionSetPathParam" required="true" type="string">actionSetPath</PARAM> |
|
95 |
</PARAMETERS> |
|
96 |
<ARCS> |
|
97 |
<ARC to="duplicateScan"/> |
|
98 |
</ARCS> |
|
99 |
</NODE> |
|
100 |
|
|
101 |
<NODE name="duplicateScan" type="SubmitHadoopJob"> |
|
102 |
<DESCRIPTION>create the raw graph</DESCRIPTION> |
|
103 |
<PARAMETERS> |
|
104 |
<PARAM required="true" type="string" name="hadoopJob" managedBy="system">executeOozieJob</PARAM> |
|
105 |
<PARAM required="true" type="string" name="cluster" managedBy="system">IIS</PARAM> |
|
106 |
<PARAM required="true" type="string" name="envParams" managedBy="system"> |
|
107 |
{ |
|
108 |
'rawSet' : 'actionSetPath', |
|
109 |
'actionSetId' : 'setActionSetId' |
|
110 |
} |
|
111 |
</PARAM> |
|
112 |
<PARAM required="true" type="string" name="params" managedBy="system"> |
|
113 |
{ |
|
114 |
'oozie.wf.application.path' : '/lib/dnet/dedup/scan/oozie_app', |
|
115 |
'graphBasePath' : '/var/lib/dnet/graph_BETA/graph_raw', |
|
116 |
'workingPath' : '/tmp/beta_dedup' |
|
117 |
} |
|
118 |
</PARAM> |
|
119 |
<PARAM required="true" type="string" name="oozieReportActionsCsv" managedBy="system">build-report</PARAM> |
|
120 |
</PARAMETERS> |
|
121 |
<ARCS> |
|
122 |
<ARC to="updateActionSets"/> |
|
123 |
</ARCS> |
|
124 |
</NODE> |
|
125 |
|
|
126 |
<NODE name="updateActionSets" type="UpdateActionSets"> |
|
127 |
<DESCRIPTION>update action sets</DESCRIPTION> |
|
128 |
<PARAMETERS/> |
|
129 |
<ARCS> |
|
130 |
<ARC to="success"/> |
|
131 |
</ARCS> |
|
132 |
</NODE> |
|
133 |
|
|
134 |
</CONFIGURATION> |
|
135 |
<STATUS> |
|
136 |
<LAST_EXECUTION_ID>wf_20200320_224915_420</LAST_EXECUTION_ID> |
|
137 |
<LAST_EXECUTION_DATE>2020-03-20T23:07:47+00:00</LAST_EXECUTION_DATE> |
|
138 |
<LAST_EXECUTION_STATUS>SUCCESS</LAST_EXECUTION_STATUS> |
|
139 |
<LAST_EXECUTION_ERROR/> |
|
140 |
<LAST_EXECUTION_OUTPUT name="mainlog:storeSimilarities:count">0</LAST_EXECUTION_OUTPUT> |
|
141 |
<LAST_EXECUTION_OUTPUT name="mainlog:storeDissimilarities:count">0</LAST_EXECUTION_OUTPUT> |
|
142 |
</STATUS> |
|
143 |
</BODY> |
|
144 |
</RESOURCE_PROFILE> |
modules/dnet-openaireplus-workflows/trunk/src/main/resources/eu/dnetlib/test/profiles/openaireplus/workflows/dedup/dedup.ocean.meta.xml | ||
---|---|---|
1 |
<RESOURCE_PROFILE> |
|
2 |
<HEADER> |
|
3 |
<RESOURCE_IDENTIFIER value="6ebfea46-0329-4d25-97eb-f06133b992bb_TWV0YVdvcmtmbG93RFNSZXNvdXJjZXMvTWV0YVdvcmtmbG93RFNSZXNvdXJjZVR5cGU="/> |
|
4 |
<RESOURCE_TYPE value="MetaWorkflowDSResourceType"/> |
|
5 |
<RESOURCE_KIND value="MetaWorkflowDSResources"/> |
|
6 |
<RESOURCE_URI value=""/> |
|
7 |
<DATE_OF_CREATION value="2020-02-18T15:01:54+00:00"/> |
|
8 |
</HEADER> |
|
9 |
<BODY> |
|
10 |
<METAWORKFLOW_NAME family="Deduplication [OCEAN]">Deduplication [OCEAN]</METAWORKFLOW_NAME> |
|
11 |
<METAWORKFLOW_DESCRIPTION/> |
|
12 |
<METAWORKFLOW_SECTION>InfoSpace Deduplication</METAWORKFLOW_SECTION> |
|
13 |
<ADMIN_EMAIL>alessia.bardi@isti.cnr.it,claudio.atzori@isti.cnr.it,michele.debonis@isti.cnr.it</ADMIN_EMAIL> |
|
14 |
<CONFIGURATION status="EXECUTABLE"> |
|
15 |
<WORKFLOW id="2483c5e4-dde8-4452-bd9c-38d436d8d941_V29ya2Zsb3dEU1Jlc291cmNlcy9Xb3JrZmxvd0RTUmVzb3VyY2VUeXBl" name="Duplicate scan"> |
|
16 |
</WORKFLOW> |
|
17 |
</CONFIGURATION> |
|
18 |
<SCHEDULING enabled="false"> |
|
19 |
<CRON>29 5 22 ? * *</CRON> |
|
20 |
<MININTERVAL>10080</MININTERVAL> |
|
21 |
</SCHEDULING> |
|
22 |
</BODY> |
|
23 |
</RESOURCE_PROFILE> |
modules/dnet-openaireplus-workflows/trunk/src/main/resources/eu/dnetlib/test/profiles/openaireplus/workflows/provision/ocean.provision.meta.xml | ||
---|---|---|
1 |
<RESOURCE_PROFILE> |
|
2 |
<HEADER> |
|
3 |
<RESOURCE_IDENTIFIER value="d511a255-6356-4116-a862-5c1cc4ebd63c_TWV0YVdvcmtmbG93RFNSZXNvdXJjZXMvTWV0YVdvcmtmbG93RFNSZXNvdXJjZVR5cGU="/> |
|
4 |
<RESOURCE_TYPE value="MetaWorkflowDSResourceType"/> |
|
5 |
<RESOURCE_KIND value="MetaWorkflowDSResources"/> |
|
6 |
<RESOURCE_URI value=""/> |
|
7 |
<DATE_OF_CREATION value="2020-02-18T15:01:54+00:00"/> |
|
8 |
</HEADER> |
|
9 |
<BODY> |
|
10 |
<METAWORKFLOW_NAME family="OpenAIRE Provision [OCEAN]">OpenAIRE Provision [OCEAN]</METAWORKFLOW_NAME> |
|
11 |
<METAWORKFLOW_DESCRIPTION/> |
|
12 |
<METAWORKFLOW_SECTION>InfoSpace Provision</METAWORKFLOW_SECTION> |
|
13 |
<ADMIN_EMAIL>alessia.bardi@isti.cnr.it,claudio.atzori@isti.cnr.it</ADMIN_EMAIL> |
|
14 |
<CONFIGURATION status="EXECUTABLE"> |
|
15 |
<WORKFLOW id="74d90d54-bea4-4a79-82d9-adddcc89e660_V29ya2Zsb3dEU1Jlc291cmNlcy9Xb3JrZmxvd0RTUmVzb3VyY2VUeXBl" name="provision"> |
|
16 |
</WORKFLOW> |
|
17 |
</CONFIGURATION> |
|
18 |
<SCHEDULING enabled="false"> |
|
19 |
<CRON>29 5 22 ? * *</CRON> |
|
20 |
<MININTERVAL>10080</MININTERVAL> |
|
21 |
</SCHEDULING> |
|
22 |
</BODY> |
|
23 |
</RESOURCE_PROFILE> |
modules/dnet-openaireplus-workflows/trunk/src/main/resources/eu/dnetlib/test/profiles/openaireplus/workflows/provision/ocean.provision.xml | ||
---|---|---|
1 |
<RESOURCE_PROFILE> |
|
2 |
<HEADER> |
|
3 |
<RESOURCE_IDENTIFIER value="74d90d54-bea4-4a79-82d9-adddcc89e660_V29ya2Zsb3dEU1Jlc291cmNlcy9Xb3JrZmxvd0RTUmVzb3VyY2VUeXBl"/> |
|
4 |
<RESOURCE_TYPE value="WorkflowDSResourceType"/> |
|
5 |
<RESOURCE_KIND value="WorkflowDSResources"/> |
|
6 |
<RESOURCE_URI value=""/> |
|
7 |
<DATE_OF_CREATION value="2014-08-01T18:13:51.0Z"/> |
|
8 |
</HEADER> |
|
9 |
<BODY> |
|
10 |
<WORKFLOW_NAME>Data Provision [OCEAN]</WORKFLOW_NAME> |
|
11 |
<WORKFLOW_TYPE>Data Provision</WORKFLOW_TYPE> |
|
12 |
<WORKFLOW_PRIORITY>30</WORKFLOW_PRIORITY> |
|
13 |
<CONFIGURATION start="manual"> |
|
14 |
|
|
15 |
<!-- PREPARE NODES --> |
|
16 |
<NODE name="setGraphBasePath" type="SetEnvParameter" isStart="true"> |
|
17 |
<DESCRIPTION>Set the target directory to materialize the graph</DESCRIPTION> |
|
18 |
<PARAMETERS> |
|
19 |
<PARAM managedBy="system" name="parameterName" required="true" type="string">graphBasePath</PARAM> |
|
20 |
<PARAM managedBy="user" name="parameterValue" required="true" type="string">/var/lib/dnet/graph_BETA</PARAM> |
|
21 |
</PARAMETERS> |
|
22 |
<ARCS> |
|
23 |
<ARC to="rawGraph"/> |
|
24 |
</ARCS> |
|
25 |
</NODE> |
|
26 |
|
|
27 |
<NODE name="rawGraph" type="SubmitHadoopJob" isJoin="true"> |
|
28 |
<DESCRIPTION>create the raw graph</DESCRIPTION> |
|
29 |
<PARAMETERS> |
|
30 |
<PARAM required="true" type="string" name="hadoopJob" managedBy="system">executeOozieJob</PARAM> |
|
31 |
<PARAM required="true" type="string" name="cluster" managedBy="system">IIS</PARAM> |
|
32 |
<PARAM required="true" type="string" name="envParams" managedBy="system"> |
|
33 |
{ |
|
34 |
'graphBasePath' : 'graphBasePath' |
|
35 |
} |
|
36 |
</PARAM> |
|
37 |
<PARAM required="true" type="string" name="params" managedBy="system"> |
|
38 |
{ |
|
39 |
'oozie.wf.application.path' : '/lib/dnet/migration/wfs/regular_all_steps/oozie_app', |
|
40 |
'mongoURL' : 'mongodb://beta.services.openaire.eu', |
|
41 |
'mongoDb' : 'mdstore', |
|
42 |
'postgresURL' : 'jdbc:postgresql://beta.services.openaire.eu:5432/dnet_openaireplus', |
|
43 |
'postgresUser' : 'dnet', |
|
44 |
'postgresPassword' : 'dnetPwd', |
|
45 |
'workingPath' : '/tmp/dhp_migration_beta', |
|
46 |
'reuseContent' : 'true' |
|
47 |
} |
|
48 |
</PARAM> |
|
49 |
<PARAM required="true" type="boolean" name="simulation" managedBy="user">false</PARAM> |
|
50 |
<PARAM required="true" type="string" name="oozieReportActionsCsv" managedBy="system">build-report</PARAM> |
|
51 |
</PARAMETERS> |
|
52 |
<ARCS> |
|
53 |
<ARC to="success"/> |
|
54 |
</ARCS> |
|
55 |
</NODE> |
|
56 |
|
|
57 |
</CONFIGURATION> |
|
58 |
<STATUS> |
|
59 |
</STATUS> |
|
60 |
</BODY> |
|
61 |
</RESOURCE_PROFILE> |
modules/dnet-openaireplus-workflows/trunk/src/main/resources/eu/dnetlib/test/profiles/openaireplus/workflows/provision/provision.ocean.meta.xml | ||
---|---|---|
1 |
<RESOURCE_PROFILE> |
|
2 |
<HEADER> |
|
3 |
<RESOURCE_IDENTIFIER value="d511a255-6356-4116-a862-5c1cc4ebd63c_TWV0YVdvcmtmbG93RFNSZXNvdXJjZXMvTWV0YVdvcmtmbG93RFNSZXNvdXJjZVR5cGU="/> |
|
4 |
<RESOURCE_TYPE value="MetaWorkflowDSResourceType"/> |
|
5 |
<RESOURCE_KIND value="MetaWorkflowDSResources"/> |
|
6 |
<RESOURCE_URI value=""/> |
|
7 |
<DATE_OF_CREATION value="2020-02-18T15:01:54+00:00"/> |
|
8 |
</HEADER> |
|
9 |
<BODY> |
|
10 |
<METAWORKFLOW_NAME family="OpenAIRE Provision [OCEAN]">OpenAIRE Provision [OCEAN]</METAWORKFLOW_NAME> |
|
11 |
<METAWORKFLOW_DESCRIPTION/> |
|
12 |
<METAWORKFLOW_SECTION>InfoSpace Provision</METAWORKFLOW_SECTION> |
|
13 |
<ADMIN_EMAIL>alessia.bardi@isti.cnr.it,claudio.atzori@isti.cnr.it</ADMIN_EMAIL> |
|
14 |
<CONFIGURATION status="EXECUTABLE"> |
|
15 |
<WORKFLOW id="74d90d54-bea4-4a79-82d9-adddcc89e660_V29ya2Zsb3dEU1Jlc291cmNlcy9Xb3JrZmxvd0RTUmVzb3VyY2VUeXBl" name="provision"> |
|
16 |
</WORKFLOW> |
|
17 |
</CONFIGURATION> |
|
18 |
<SCHEDULING enabled="false"> |
|
19 |
<CRON>29 5 22 ? * *</CRON> |
|
20 |
<MININTERVAL>10080</MININTERVAL> |
|
21 |
</SCHEDULING> |
|
22 |
</BODY> |
|
23 |
</RESOURCE_PROFILE> |
modules/dnet-openaireplus-workflows/trunk/src/main/resources/eu/dnetlib/test/profiles/openaireplus/workflows/provision/provision.ocean.xml | ||
---|---|---|
1 |
<RESOURCE_PROFILE> |
|
2 |
<HEADER> |
|
3 |
<RESOURCE_IDENTIFIER value="74d90d54-bea4-4a79-82d9-adddcc89e660_V29ya2Zsb3dEU1Jlc291cmNlcy9Xb3JrZmxvd0RTUmVzb3VyY2VUeXBl"/> |
|
4 |
<RESOURCE_TYPE value="WorkflowDSResourceType"/> |
|
5 |
<RESOURCE_KIND value="WorkflowDSResources"/> |
|
6 |
<RESOURCE_URI value=""/> |
|
7 |
<DATE_OF_CREATION value="2014-08-01T18:13:51.0Z"/> |
|
8 |
</HEADER> |
|
9 |
<BODY> |
|
10 |
<WORKFLOW_NAME>Data Provision [OCEAN]</WORKFLOW_NAME> |
|
11 |
<WORKFLOW_TYPE>Data Provision</WORKFLOW_TYPE> |
|
12 |
<WORKFLOW_PRIORITY>30</WORKFLOW_PRIORITY> |
|
13 |
<CONFIGURATION start="manual"> |
|
14 |
|
|
15 |
<!-- PREPARE NODES --> |
|
16 |
<NODE name="setGraphBasePath" type="SetEnvParameter" isStart="true"> |
|
17 |
<DESCRIPTION>Set the target directory to materialize the graph</DESCRIPTION> |
|
18 |
<PARAMETERS> |
|
19 |
<PARAM managedBy="system" name="parameterName" required="true" type="string">graphBasePath</PARAM> |
|
20 |
<PARAM managedBy="user" name="parameterValue" required="true" type="string">/var/lib/dnet/graph_BETA</PARAM> |
|
21 |
</PARAMETERS> |
|
22 |
<ARCS> |
|
23 |
<ARC to="rawGraph"/> |
|
24 |
</ARCS> |
|
25 |
</NODE> |
|
26 |
|
|
27 |
<NODE name="rawGraph" type="SubmitHadoopJob" isJoin="true"> |
|
28 |
<DESCRIPTION>create the raw graph</DESCRIPTION> |
|
29 |
<PARAMETERS> |
|
30 |
<PARAM required="true" type="string" name="hadoopJob" managedBy="system">executeOozieJob</PARAM> |
|
31 |
<PARAM required="true" type="string" name="cluster" managedBy="system">IIS</PARAM> |
|
32 |
<PARAM required="true" type="string" name="envParams" managedBy="system"> |
|
33 |
{ |
|
34 |
'graphBasePath' : 'graphBasePath' |
|
35 |
} |
|
36 |
</PARAM> |
|
37 |
<PARAM required="true" type="string" name="params" managedBy="system"> |
|
38 |
{ |
|
39 |
'oozie.wf.application.path' : '/lib/dnet/migration/wfs/regular_all_steps/oozie_app', |
|
40 |
'mongoURL' : 'mongodb://beta.services.openaire.eu', |
|
41 |
'mongoDb' : 'mdstore', |
|
42 |
'postgresURL' : 'jdbc:postgresql://beta.services.openaire.eu:5432/dnet_openaireplus', |
|
43 |
'postgresUser' : 'dnet', |
|
44 |
'postgresPassword' : 'dnetPwd', |
|
45 |
'workingPath' : '/tmp/dhp_migration_beta', |
|
46 |
'reuseContent' : 'true' |
|
47 |
} |
|
48 |
</PARAM> |
|
49 |
<PARAM required="true" type="string" name="oozieReportActionsCsv" managedBy="system">build-report</PARAM> |
|
50 |
</PARAMETERS> |
|
51 |
<ARCS> |
|
52 |
<ARC to="setRawGraphPath"/> |
|
53 |
</ARCS> |
|
54 |
</NODE> |
|
55 |
|
|
56 |
<NODE name="setRawGraphPath" type="SetEnvParameter"> |
|
57 |
<DESCRIPTION>Set the path containing the raw graph</DESCRIPTION> |
|
58 |
<PARAMETERS> |
|
59 |
<PARAM managedBy="system" name="parameterName" required="true" type="string">rawGraphPath</PARAM> |
|
60 |
<PARAM managedBy="user" name="parameterValue" required="true" type="string">/var/lib/dnet/graph_BETA/graph_raw</PARAM> |
|
61 |
</PARAMETERS> |
|
62 |
<ARCS> |
|
63 |
<ARC to="setDedupWorkingPath"/> |
|
64 |
</ARCS> |
|
65 |
</NODE> |
|
66 |
|
|
67 |
<NODE name="setDedupWorkingPath" type="SetEnvParameter"> |
|
68 |
<DESCRIPTION>Set the path used by dedup workflows to store intermediate data</DESCRIPTION> |
|
69 |
<PARAMETERS> |
|
70 |
<PARAM managedBy="system" name="parameterName" required="true" type="string">dedupWorkingPath</PARAM> |
|
71 |
<PARAM managedBy="user" name="parameterValue" required="true" type="string">/tmp/beta_dedup</PARAM> |
|
72 |
</PARAMETERS> |
|
73 |
<ARCS> |
|
74 |
<ARC to="setDedupGraphPath"/> |
|
75 |
</ARCS> |
|
76 |
</NODE> |
|
77 |
|
|
78 |
<NODE name="setDedupGraphPath" type="SetEnvParameter"> |
|
79 |
<DESCRIPTION>Set the target path to store the deduped graph</DESCRIPTION> |
|
80 |
<PARAMETERS> |
|
81 |
<PARAM managedBy="system" name="parameterName" required="true" type="string">dedupGraphPath</PARAM> |
|
82 |
<PARAM managedBy="user" name="parameterValue" required="true" type="string">/var/lib/dnet/graph_BETA/graph_dedup</PARAM> |
|
83 |
</PARAMETERS> |
|
84 |
<ARCS> |
|
85 |
<ARC to="dedupOrganization"/> |
|
86 |
</ARCS> |
|
87 |
</NODE> |
|
88 |
|
|
89 |
<NODE name="dedupOrganization" type="SubmitHadoopJob"> |
|
90 |
<DESCRIPTION>deduplicate organizations</DESCRIPTION> |
|
91 |
<PARAMETERS> |
|
92 |
<PARAM required="true" type="string" name="hadoopJob" managedBy="system">executeOozieJob</PARAM> |
|
93 |
<PARAM required="true" type="string" name="cluster" managedBy="system">IIS</PARAM> |
|
94 |
<PARAM required="true" type="string" name="envParams" managedBy="system"> |
|
95 |
{ |
|
96 |
'graphBasePath' : 'rawGraphPath', |
|
97 |
'workingPath' : 'dedupWorkingPath', |
|
98 |
'dedupGraphPath': 'dedupGraphPath' |
|
99 |
} |
|
100 |
</PARAM> |
|
101 |
<PARAM required="true" type="string" name="params" managedBy="system"> |
|
102 |
{ |
|
103 |
'oozie.wf.application.path' : '/lib/dnet/dedup/scan/oozie_app', |
|
104 |
'actionSetId' : 'dedup-similarity-organization' |
|
105 |
} |
|
106 |
</PARAM> |
|
107 |
<PARAM required="true" type="string" name="oozieReportActionsCsv" managedBy="system">build-report</PARAM> |
|
108 |
</PARAMETERS> |
|
109 |
<ARCS> |
|
110 |
<ARC to="dedupResult"/> |
|
111 |
</ARCS> |
|
112 |
</NODE> |
|
113 |
|
|
114 |
<NODE name="dedupResult" type="SubmitHadoopJob"> |
|
115 |
<DESCRIPTION>deduplicate results</DESCRIPTION> |
|
116 |
<PARAMETERS> |
|
117 |
<PARAM required="true" type="string" name="hadoopJob" managedBy="system">executeOozieJob</PARAM> |
|
118 |
<PARAM required="true" type="string" name="cluster" managedBy="system">IIS</PARAM> |
|
119 |
<PARAM required="true" type="string" name="envParams" managedBy="system"> |
|
120 |
{ |
|
121 |
'graphBasePath' : 'rawGraphPath', |
|
122 |
'workingPath' : 'dedupWorkingPath', |
|
123 |
'dedupGraphPath': 'dedupGraphPath' |
|
124 |
} |
|
125 |
</PARAM> |
|
126 |
<PARAM required="true" type="string" name="params" managedBy="system"> |
|
127 |
{ |
|
128 |
'oozie.wf.application.path' : '/lib/dnet/dedup/scan/oozie_app', |
|
129 |
'actionSetId' : 'dedup-similarity-result' |
|
130 |
} |
|
131 |
</PARAM> |
|
132 |
<PARAM required="true" type="string" name="oozieReportActionsCsv" managedBy="system">build-report</PARAM> |
|
133 |
</PARAMETERS> |
|
134 |
<ARCS> |
|
135 |
<ARC to="dedupConsistency"/> |
|
136 |
</ARCS> |
|
137 |
</NODE> |
|
138 |
|
|
139 |
<NODE name="dedupConsistency" type="SubmitHadoopJob"> |
|
140 |
<DESCRIPTION>mark duplicates as deleted and redistribute the relationships</DESCRIPTION> |
|
141 |
<PARAMETERS> |
|
142 |
<PARAM required="true" type="string" name="hadoopJob" managedBy="system">executeOozieJob</PARAM> |
|
143 |
<PARAM required="true" type="string" name="cluster" managedBy="system">IIS</PARAM> |
|
144 |
<PARAM required="true" type="string" name="envParams" managedBy="system"> |
|
145 |
{ |
|
146 |
'graphBasePath' : 'rawGraphPath', |
|
147 |
'workingPath' : 'dedupWorkingPath', |
|
148 |
'dedupGraphPath': 'dedupGraphPath' |
|
149 |
} |
|
150 |
</PARAM> |
|
151 |
<PARAM required="true" type="string" name="params" managedBy="system"> |
|
152 |
{ |
|
153 |
'oozie.wf.application.path' : '/lib/dnet/dedup/consistency/oozie_app' |
|
154 |
} |
|
155 |
</PARAM> |
|
156 |
<PARAM required="true" type="string" name="oozieReportActionsCsv" managedBy="system">build-report</PARAM> |
|
157 |
</PARAMETERS> |
|
158 |
<ARCS> |
|
159 |
<ARC to="success"/> |
|
160 |
</ARCS> |
|
161 |
</NODE> |
|
162 |
|
|
163 |
</CONFIGURATION> |
|
164 |
<STATUS> |
|
165 |
</STATUS> |
|
166 |
</BODY> |
|
167 |
</RESOURCE_PROFILE> |
Also available in: Unified diff
WIP: defining data provision workflow to be run on the OCEAN cluster