1
|
<?xml version="1.0" encoding="UTF-8"?>
|
2
|
<RESOURCE_PROFILE xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance">
|
3
|
<HEADER>
|
4
|
<RESOURCE_IDENTIFIER value="ae261797-c373-48ce-8581-f1b66f702323_V29ya2Zsb3dEU1Jlc291cmNlcy9Xb3JrZmxvd0RTUmVzb3VyY2VUeXBl"/>
|
5
|
<RESOURCE_TYPE value="WorkflowDSResourceType"/>
|
6
|
<RESOURCE_KIND value="WorkflowDSResources"/>
|
7
|
<RESOURCE_URI value=""/>
|
8
|
<DATE_OF_CREATION value="2006-05-04T18:13:51.0Z"/>
|
9
|
</HEADER>
|
10
|
<BODY>
|
11
|
<WORKFLOW_NAME>Offline Deduplication</WORKFLOW_NAME>
|
12
|
<WORKFLOW_TYPE>Deduplication</WORKFLOW_TYPE>
|
13
|
<WORKFLOW_PRIORITY>30</WORKFLOW_PRIORITY>
|
14
|
<CONFIGURATION start="manual">
|
15
|
|
16
|
<NODE name="hadoopConfig" type="SetClusterAndTable" isStart="true">
|
17
|
<DESCRIPTION>Set table name</DESCRIPTION>
|
18
|
<PARAMETERS>
|
19
|
<PARAM required="true" type="string" name="cluster" managedBy="system">DM</PARAM>
|
20
|
<PARAM required="true" type="string" name="tableParam" managedBy="system">tableName</PARAM>
|
21
|
</PARAMETERS>
|
22
|
<ARCS>
|
23
|
<ARC to="cleanupMap" />
|
24
|
</ARCS>
|
25
|
</NODE>
|
26
|
<NODE name="setPath" type="SetHdfsPathJob" isStart="true">
|
27
|
<DESCRIPTION>Set map path name</DESCRIPTION>
|
28
|
<PARAMETERS>
|
29
|
<PARAM required="true" type="string" name="path" managedBy="user">/tmp/anchorMap</PARAM>
|
30
|
<PARAM required="true" type="string" name="pathParam" managedBy="system">mapPath</PARAM>
|
31
|
</PARAMETERS>
|
32
|
<ARCS>
|
33
|
<ARC to="cleanupMap" />
|
34
|
</ARCS>
|
35
|
</NODE>
|
36
|
|
37
|
<!-- postprocess -->
|
38
|
|
39
|
<NODE name="cleanupMap" type="DeleteHdfsPathJob" isJoin="true">
|
40
|
<DESCRIPTION>hdfs cleanup (xml)</DESCRIPTION>
|
41
|
<PARAMETERS>
|
42
|
<PARAM required="true" type="string" name="cluster" managedBy="system">DM</PARAM>
|
43
|
<PARAM required="true" type="string" name="envParams" managedBy="system">
|
44
|
{
|
45
|
'path' : 'mapPath'
|
46
|
}
|
47
|
</PARAM>
|
48
|
</PARAMETERS>
|
49
|
<ARCS>
|
50
|
<ARC to="buildAnchorMap" />
|
51
|
</ARCS>
|
52
|
</NODE>
|
53
|
|
54
|
<NODE name="buildAnchorMap" type="SubmitHadoopJob">
|
55
|
<DESCRIPTION>redirect rels</DESCRIPTION>
|
56
|
<PARAMETERS>
|
57
|
<PARAM required="true" type="string" name="hadoopJob" managedBy="system">buildMergedToAnchorMapJob</PARAM>
|
58
|
<PARAM required="true" type="string" name="envParams" managedBy="system">
|
59
|
{
|
60
|
'cluster' : 'cluster',
|
61
|
'hbase.mapred.inputtable' : 'tableName',
|
62
|
'hbase.mapreduce.inputtable' : 'tableName',
|
63
|
'hbase.mapred.outputtable' : 'tableName',
|
64
|
'hbase.mapreduce.outputtable' : 'tableName',
|
65
|
'mapred.output.dir' : 'mapPath'
|
66
|
}
|
67
|
</PARAM>
|
68
|
</PARAMETERS>
|
69
|
<ARCS>
|
70
|
<ARC to="success" />
|
71
|
</ARCS>
|
72
|
</NODE>
|
73
|
|
74
|
|
75
|
</CONFIGURATION>
|
76
|
<STATUS />
|
77
|
</BODY>
|
78
|
</RESOURCE_PROFILE>
|