1
|
<RESOURCE_PROFILE>
|
2
|
<HEADER>
|
3
|
<RESOURCE_IDENTIFIER value="e03f256e-1e4d-4b3d-9c07-91faf5d25208_V29ya2Zsb3dEU1Jlc291cmNlcy9Xb3JrZmxvd0RTUmVzb3VyY2VUeXBl"/>
|
4
|
<RESOURCE_TYPE value="WorkflowDSResourceType"/>
|
5
|
<RESOURCE_KIND value="WorkflowDSResources"/>
|
6
|
<RESOURCE_URI value=""/>
|
7
|
<DATE_OF_CREATION value="2021-04-19T13:28:18+00:00"/>
|
8
|
</HEADER>
|
9
|
<BODY>
|
10
|
<WORKFLOW_NAME>Import DOIboost</WORKFLOW_NAME>
|
11
|
<WORKFLOW_TYPE>Import InfoSpace</WORKFLOW_TYPE>
|
12
|
<WORKFLOW_PRIORITY>30</WORKFLOW_PRIORITY>
|
13
|
<CONFIGURATION start="manual">
|
14
|
<NODE isStart="true" name="setMAGDumpPath" type="SetEnvParameter">
|
15
|
<DESCRIPTION>set the input path for MAG</DESCRIPTION>
|
16
|
<PARAMETERS>
|
17
|
<PARAM managedBy="system" name="parameterName" required="true" type="string">MAGDumpPath</PARAM>
|
18
|
<PARAM managedBy="user" name="parameterValue" required="true" type="string">/data/doiboost/mag-2020-11-09</PARAM>
|
19
|
</PARAMETERS>
|
20
|
<ARCS>
|
21
|
<ARC to="waitConfig"/>
|
22
|
</ARCS>
|
23
|
</NODE>
|
24
|
<NODE isStart="true" name="setInputPathMAG" type="SetEnvParameter">
|
25
|
<DESCRIPTION>set the input path for MAG</DESCRIPTION>
|
26
|
<PARAMETERS>
|
27
|
<PARAM managedBy="system" name="parameterName" required="true" type="string">inputPathMAG</PARAM>
|
28
|
<PARAM managedBy="user" name="parameterValue" required="true" type="string">/data/doiboost/input/mag</PARAM>
|
29
|
</PARAMETERS>
|
30
|
<ARCS>
|
31
|
<ARC to="waitConfig"/>
|
32
|
</ARCS>
|
33
|
</NODE>
|
34
|
<NODE isStart="true" name="setInputPathCrossref" type="SetEnvParameter">
|
35
|
<DESCRIPTION>set the input path for Crossref</DESCRIPTION>
|
36
|
<PARAMETERS>
|
37
|
<PARAM managedBy="system" name="parameterName" required="true" type="string">inputPathCrossref</PARAM>
|
38
|
<PARAM managedBy="user" name="parameterValue" required="true" type="string">/data/doiboost/input/crossref</PARAM>
|
39
|
</PARAMETERS>
|
40
|
<ARCS>
|
41
|
<ARC to="waitConfig"/>
|
42
|
</ARCS>
|
43
|
</NODE>
|
44
|
<NODE isStart="true" name="setCrossrefTimestamp" type="SetEnvParameter">
|
45
|
<DESCRIPTION>set the timestamp for the Crossref incremental harvesting</DESCRIPTION>
|
46
|
<PARAMETERS>
|
47
|
<PARAM managedBy="system" name="parameterName" required="true" type="string">crossrefTimestamp</PARAM>
|
48
|
<PARAM managedBy="user" name="parameterValue" required="false" type="string">1607614921429</PARAM>
|
49
|
</PARAMETERS>
|
50
|
<ARCS>
|
51
|
<ARC to="waitConfig"/>
|
52
|
</ARCS>
|
53
|
</NODE>
|
54
|
<NODE isStart="true" name="setInputPathUnpayWall" type="SetEnvParameter">
|
55
|
<DESCRIPTION>set the input path for UnpayWall</DESCRIPTION>
|
56
|
<PARAMETERS>
|
57
|
<PARAM managedBy="system" name="parameterName" required="true" type="string">inputPathUnpayWall</PARAM>
|
58
|
<PARAM managedBy="user" name="parameterValue" required="true" type="string">/data/doiboost/input/unpayWall</PARAM>
|
59
|
</PARAMETERS>
|
60
|
<ARCS>
|
61
|
<ARC to="waitConfig"/>
|
62
|
</ARCS>
|
63
|
</NODE>
|
64
|
<NODE isStart="true" name="setInputPathOrcid" type="SetEnvParameter">
|
65
|
<DESCRIPTION>set the input path for ORCID</DESCRIPTION>
|
66
|
<PARAMETERS>
|
67
|
<PARAM managedBy="system" name="parameterName" required="true" type="string">inputPathOrcid</PARAM>
|
68
|
<PARAM managedBy="user" name="parameterValue" required="true" type="string">/data/orcid_activities_2020/last_orcid_dataset</PARAM>
|
69
|
</PARAMETERS>
|
70
|
<ARCS>
|
71
|
<ARC to="waitConfig"/>
|
72
|
</ARCS>
|
73
|
</NODE>
|
74
|
<NODE isStart="true" name="setWorkingPathOrcid" type="SetEnvParameter">
|
75
|
<DESCRIPTION>set the working path for ORCID</DESCRIPTION>
|
76
|
<PARAMETERS>
|
77
|
<PARAM managedBy="system" name="parameterName" required="true" type="string">workingPathOrcid</PARAM>
|
78
|
<PARAM managedBy="user" name="parameterValue" required="true" type="string">/data/doiboost/input/orcid</PARAM>
|
79
|
</PARAMETERS>
|
80
|
<ARCS>
|
81
|
<ARC to="waitConfig"/>
|
82
|
</ARCS>
|
83
|
</NODE>
|
84
|
<NODE isStart="true" name="setHostedByMapPath" type="SetEnvParameter">
|
85
|
<DESCRIPTION>set the hostedBy map path</DESCRIPTION>
|
86
|
<PARAMETERS>
|
87
|
<PARAM managedBy="system" name="parameterName" required="true" type="string">hostedByMapPath</PARAM>
|
88
|
<PARAM managedBy="user" name="parameterValue" required="true" type="string">/data/doiboost/input/hostedBy/hbMap.gz</PARAM>
|
89
|
</PARAMETERS>
|
90
|
<ARCS>
|
91
|
<ARC to="waitConfig"/>
|
92
|
</ARCS>
|
93
|
</NODE>
|
94
|
<NODE isStart="true" name="setResumeFrom" type="SetEnvParameter">
|
95
|
<DESCRIPTION>set the oozie workflow name from which the execution will be resumed</DESCRIPTION>
|
96
|
<PARAMETERS>
|
97
|
<PARAM managedBy="system" name="parameterName" required="true" type="string">resumeFrom</PARAM>
|
98
|
<PARAM function="validValues(['ConvertCrossrefToOAF','ResetMagWorkingPath', 'PreprocessMag', 'PreprocessUW', 'PreprocessORCID', 'CreateDOIBoost', 'GenerateActionSet'])" managedBy="user" name="parameterValue" required="false" type="string">PreprocessORCID</PARAM>
|
99
|
</PARAMETERS>
|
100
|
<ARCS>
|
101
|
<ARC to="waitConfig"/>
|
102
|
</ARCS>
|
103
|
</NODE>
|
104
|
<NODE isJoin="true" name="waitConfig">
|
105
|
<DESCRIPTION>wait configurations</DESCRIPTION>
|
106
|
<PARAMETERS/>
|
107
|
<ARCS>
|
108
|
<ARC to="prepareActionSets"/>
|
109
|
</ARCS>
|
110
|
</NODE>
|
111
|
<NODE name="prepareActionSets" type="PrepareActionSets">
|
112
|
<DESCRIPTION>prepare action sets</DESCRIPTION>
|
113
|
<PARAMETERS>
|
114
|
<PARAM managedBy="system" name="sets" required="true" type="string">
|
115
|
[
|
116
|
{
|
117
|
'set' : 'doiboost',
|
118
|
'jobProperty' : 'export_action_set_doiboost',
|
119
|
'enablingProperty' : 'active_doiboost',
|
120
|
'enabled' : 'true'
|
121
|
}
|
122
|
]
|
123
|
</PARAM>
|
124
|
</PARAMETERS>
|
125
|
<ARCS>
|
126
|
<ARC to="extractOutputPath"/>
|
127
|
</ARCS>
|
128
|
</NODE>
|
129
|
<NODE name="extractOutputPath" type="ExtractOutputPath">
|
130
|
<DESCRIPTION>extract the hdfs output path generated in the previous node</DESCRIPTION>
|
131
|
<PARAMETERS>
|
132
|
<PARAM managedBy="system" name="hdfsOutputPathParam" required="true" type="string">outputPath</PARAM>
|
133
|
</PARAMETERS>
|
134
|
<ARCS>
|
135
|
<ARC to="updateDOIBoost"/>
|
136
|
</ARCS>
|
137
|
</NODE>
|
138
|
<NODE name="updateDOIBoost" type="SubmitHadoopJob">
|
139
|
<DESCRIPTION>prepare a new version of DOIBoost</DESCRIPTION>
|
140
|
<PARAMETERS>
|
141
|
<PARAM managedBy="system" name="hadoopJob" required="true" type="string">executeOozieJob</PARAM>
|
142
|
<PARAM managedBy="system" name="cluster" required="true" type="string">IIS</PARAM>
|
143
|
<PARAM managedBy="system" name="envParams" required="true" type="string">
|
144
|
{
|
145
|
'crossrefTimestamp' : 'crossrefTimestamp',
|
146
|
'hostedByMapPath' : 'hostedByMapPath',
|
147
|
'MAGDumpPath' :'MAGDumpPath',
|
148
|
'inputPathMAG' : 'inputPathMAG',
|
149
|
'inputPathCrossref' : 'inputPathCrossref',
|
150
|
'inputPathUnpayWall' : 'inputPathUnpayWall',
|
151
|
'inputPathOrcid' : 'inputPathOrcid',
|
152
|
'outputPath' : 'outputPath',
|
153
|
'workingPathOrcid':'workingPathOrcid',
|
154
|
'resumeFrom' : 'resumeFrom'
|
155
|
}
|
156
|
</PARAM>
|
157
|
<PARAM managedBy="system" name="params" required="true" type="string">
|
158
|
{
|
159
|
'oozie.wf.application.path' : ' /lib/dnet/STABLE_IDS/actionmanager/doiboost/oozie_app',
|
160
|
'workingPath' : '/data/doiboost/process',
|
161
|
'sparkExecutorIntersectionMemory' : '9G',
|
162
|
'sparkExecutorMemory' : '6G',
|
163
|
'esServer' : 'ip-90-147-167-25.ct1.garrservices.it',
|
164
|
'esIndex' : 'crossref'
|
165
|
}
|
166
|
</PARAM>
|
167
|
<PARAM managedBy="system" name="oozieReportActionsCsv" required="true" type="string">build-report</PARAM>
|
168
|
</PARAMETERS>
|
169
|
<ARCS>
|
170
|
<ARC to="updateActionSets"/>
|
171
|
</ARCS>
|
172
|
</NODE>
|
173
|
<NODE name="updateActionSets" type="UpdateActionSets">
|
174
|
<DESCRIPTION>update action sets</DESCRIPTION>
|
175
|
<PARAMETERS/>
|
176
|
<ARCS>
|
177
|
<ARC to="success"/>
|
178
|
</ARCS>
|
179
|
</NODE>
|
180
|
</CONFIGURATION>
|
181
|
<STATUS>
|
182
|
<LAST_EXECUTION_ID>wf_20210416_122501_878</LAST_EXECUTION_ID>
|
183
|
<LAST_EXECUTION_DATE>2021-04-16T13:58:50+00:00</LAST_EXECUTION_DATE>
|
184
|
<LAST_EXECUTION_STATUS>SUCCESS</LAST_EXECUTION_STATUS>
|
185
|
<LAST_EXECUTION_ERROR/>
|
186
|
</STATUS>
|
187
|
</BODY>
|
188
|
</RESOURCE_PROFILE>
|