|
1
|
<RESOURCE_PROFILE>
|
|
2
|
<HEADER>
|
|
3
|
<RESOURCE_IDENTIFIER value="e03f256e-1e4d-4b3d-9c07-91faf5d25207_V29ya2Zsb3dEU1Jlc291cmNlcy9Xb3JrZmxvd0RTUmVzb3VyY2VUeXBl"/>
|
|
4
|
<RESOURCE_TYPE value="WorkflowDSResourceType"/>
|
|
5
|
<RESOURCE_KIND value="WorkflowDSResources"/>
|
|
6
|
<RESOURCE_URI value=""/>
|
|
7
|
<DATE_OF_CREATION value="2019-09-26T16:09:43+00:00"/>
|
|
8
|
</HEADER>
|
|
9
|
<BODY>
|
|
10
|
<WORKFLOW_NAME>Import ScholExplorer Links and entities</WORKFLOW_NAME>
|
|
11
|
<WORKFLOW_TYPE>Import InfoSpace</WORKFLOW_TYPE>
|
|
12
|
<WORKFLOW_PRIORITY>30</WORKFLOW_PRIORITY>
|
|
13
|
<CONFIGURATION start="manual">
|
|
14
|
<NODE isStart="true" name="setInputPath" type="SetHdfsFile">
|
|
15
|
<DESCRIPTION>set the hdfs output path</DESCRIPTION>
|
|
16
|
<PARAMETERS>
|
|
17
|
<PARAM managedBy="user" name="hdfsPath" required="true" type="string">/usr/dnet/data/scholixDumpExport_2019_09_26</PARAM>
|
|
18
|
<PARAM managedBy="system" name="hdfsPathParam" required="true" type="string">inputPath</PARAM>
|
|
19
|
</PARAMETERS>
|
|
20
|
<ARCS>
|
|
21
|
<ARC to="prepareActionSets"/>
|
|
22
|
</ARCS>
|
|
23
|
</NODE>
|
|
24
|
<NODE name="prepareActionSets" type="PrepareActionSets">
|
|
25
|
<DESCRIPTION>prepare action sets</DESCRIPTION>
|
|
26
|
<PARAMETERS>
|
|
27
|
<PARAM managedBy="system" name="sets" required="true" type="string">
|
|
28
|
[
|
|
29
|
{
|
|
30
|
'set' : 'scholexplorer-dump',
|
|
31
|
'jobProperty' : 'export_action_set_scholexplorer_dump',
|
|
32
|
'enablingProperty' : 'active_scholexplorer_dump',
|
|
33
|
'enabled' : 'true'
|
|
34
|
}
|
|
35
|
]
|
|
36
|
</PARAM>
|
|
37
|
</PARAMETERS>
|
|
38
|
<ARCS>
|
|
39
|
<ARC to="extractOutputPath"/>
|
|
40
|
</ARCS>
|
|
41
|
</NODE>
|
|
42
|
<NODE name="extractOutputPath" type="ExtractOutputPath">
|
|
43
|
<DESCRIPTION>extract the hdfs output path generated in the previous node</DESCRIPTION>
|
|
44
|
<PARAMETERS>
|
|
45
|
<PARAM managedBy="system" name="hdfsOutputPathParam" required="true" type="string">outputPath</PARAM>
|
|
46
|
</PARAMETERS>
|
|
47
|
<ARCS>
|
|
48
|
<ARC to="importActionSet"/>
|
|
49
|
</ARCS>
|
|
50
|
</NODE>
|
|
51
|
<NODE isJoin="true" name="importActionSet" type="SubmitHadoopJob">
|
|
52
|
<DESCRIPTION>IIS main</DESCRIPTION>
|
|
53
|
<PARAMETERS>
|
|
54
|
<PARAM managedBy="system" name="hadoopJob" required="true" type="string">importScholexplorerJob</PARAM>
|
|
55
|
<PARAM managedBy="system" name="cluster" required="true" type="string">DM</PARAM>
|
|
56
|
<PARAM managedBy="system" name="envParams" required="true" type="string">
|
|
57
|
{
|
|
58
|
'mapred.input.dir':'inputPath',
|
|
59
|
'mapred.output.dir':'outputPath'
|
|
60
|
}
|
|
61
|
</PARAM>
|
|
62
|
<PARAM managedBy="user" name="simulation" required="true" type="boolean">false</PARAM>
|
|
63
|
</PARAMETERS>
|
|
64
|
<ARCS>
|
|
65
|
<ARC to="updateActionSets"/>
|
|
66
|
</ARCS>
|
|
67
|
</NODE>
|
|
68
|
<NODE name="updateActionSets" type="UpdateActionSets">
|
|
69
|
<DESCRIPTION>update action sets</DESCRIPTION>
|
|
70
|
<PARAMETERS/>
|
|
71
|
<ARCS>
|
|
72
|
<ARC to="success"/>
|
|
73
|
</ARCS>
|
|
74
|
</NODE>
|
|
75
|
</CONFIGURATION>
|
|
76
|
<STATUS>
|
|
77
|
<LAST_EXECUTION_ID>wf_20191022_141741_170</LAST_EXECUTION_ID>
|
|
78
|
<LAST_EXECUTION_DATE>2019-10-22T14:32:03+00:00</LAST_EXECUTION_DATE>
|
|
79
|
<LAST_EXECUTION_STATUS>SUCCESS</LAST_EXECUTION_STATUS>
|
|
80
|
<LAST_EXECUTION_ERROR/>
|
|
81
|
</STATUS>
|
|
82
|
</BODY>
|
|
83
|
</RESOURCE_PROFILE>
|