|
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="551902ae-7f01-401a-a714-9b16ae5ae37e_V29ya2Zsb3dEU1Jlc291cmNlcy9Xb3JrZmxvd0RTUmVzb3VyY2VUeXBl" />
|
|
5
|
<RESOURCE_TYPE value="WorkflowDSResourceType" />
|
|
6
|
<RESOURCE_KIND value="WorkflowDSResources" />
|
|
7
|
<RESOURCE_URI value="" />
|
|
8
|
<DATE_OF_CREATION value="2014-08-01T18:13:51.0Z" />
|
|
9
|
</HEADER>
|
|
10
|
<BODY>
|
|
11
|
<WORKFLOW_NAME>Update dedup Index</WORKFLOW_NAME>
|
|
12
|
<WORKFLOW_TYPE>Deduplication</WORKFLOW_TYPE>
|
|
13
|
<WORKFLOW_PRIORITY>30</WORKFLOW_PRIORITY>
|
|
14
|
<CONFIGURATION start="manual">
|
|
15
|
|
|
16
|
<NODE name="findIndex" type="FindIndex" isStart="true">
|
|
17
|
<DESCRIPTION />
|
|
18
|
<PARAMETERS>
|
|
19
|
<PARAM name="mdFormat" type="string" managedBy="user" required="true"></PARAM>
|
|
20
|
<PARAM name="layout" type="string" managedBy="system" required="true">index</PARAM>
|
|
21
|
<PARAM name="interpretation" type="string" managedBy="system" required="true">dedup</PARAM>
|
|
22
|
</PARAMETERS>
|
|
23
|
<ARCS>
|
|
24
|
<ARC name="found" to="prepare" />
|
|
25
|
<ARC name="notFound" to="createIndex" />
|
|
26
|
</ARCS>
|
|
27
|
</NODE>
|
|
28
|
|
|
29
|
<NODE name="createIndex" type="CreateIndex">
|
|
30
|
<DESCRIPTION />
|
|
31
|
<PARAMETERS />
|
|
32
|
<ARCS>
|
|
33
|
<ARC to="prepare" />
|
|
34
|
</ARCS>
|
|
35
|
</NODE>
|
|
36
|
|
|
37
|
<NODE name="prepare" type="PrepareDedupIndexJob">
|
|
38
|
<DESCRIPTION>Prepare indexing</DESCRIPTION>
|
|
39
|
<PARAMETERS>
|
|
40
|
<PARAM name="dedupConfig" function="obtainValues('dedupOrchestrations', {})" type="string" required="true" managedBy="user"></PARAM>
|
|
41
|
<PARAM name="rottenRecordsPathParam" type="string" required="true" managedBy="system">rottenRecordsPath</PARAM>
|
|
42
|
</PARAMETERS>
|
|
43
|
<ARCS>
|
|
44
|
<ARC to="cleanupRotten" />
|
|
45
|
</ARCS>
|
|
46
|
</NODE>
|
|
47
|
|
|
48
|
<NODE name="cleanupRotten" type="DeleteHdfsPathJob">
|
|
49
|
<DESCRIPTION>hdfs cleanup (rotten)</DESCRIPTION>
|
|
50
|
<PARAMETERS>
|
|
51
|
<PARAM required="true" type="string" name="cluster" managedBy="system">DM</PARAM>
|
|
52
|
<PARAM required="true" type="string" name="envParams" managedBy="system">
|
|
53
|
{
|
|
54
|
'path' : 'rottenRecordsPath'
|
|
55
|
}
|
|
56
|
</PARAM>
|
|
57
|
</PARAMETERS>
|
|
58
|
<ARCS>
|
|
59
|
<ARC to="updateIndex" />
|
|
60
|
</ARCS>
|
|
61
|
</NODE>
|
|
62
|
|
|
63
|
<NODE name="updateIndex" type="SubmitHadoopJob">
|
|
64
|
<DESCRIPTION>M/O index records</DESCRIPTION>
|
|
65
|
<PARAMETERS>
|
|
66
|
<PARAM required="true" type="string" name="cluster" managedBy="system">DM</PARAM>
|
|
67
|
<PARAM required="true" type="string" name="hadoopJob" managedBy="system">dedupIndexFeedJob</PARAM>
|
|
68
|
<PARAM required="true" type="string" name="envParams" managedBy="system">
|
|
69
|
{
|
|
70
|
'mapred.output.dir' : 'rottenRecordsPath',
|
|
71
|
'index.fields' : 'index.fields',
|
|
72
|
'index.solr.url' : 'index.solr.url',
|
|
73
|
'index.solr.collection' : 'index.solr.collection',
|
|
74
|
'index.buffer.flush.threshold' : 'index.buffer.flush.threshold',
|
|
75
|
'index.shutdown.wait.time' : 'index.shutdown.wait.time',
|
|
76
|
'index.solr.sim.mode' : 'index.solr.sim.mode',
|
|
77
|
'index.feed.timestamp' : 'index.feed.timestamp',
|
|
78
|
'entityTypeId' : 'entityTypeId',
|
|
79
|
'entityType' : 'entityType',
|
|
80
|
'actionset' : 'actionset'
|
|
81
|
}
|
|
82
|
</PARAM>
|
|
83
|
<PARAM required="true" type="string" name="sysParams" managedBy="system">
|
|
84
|
{
|
|
85
|
'hbase.mapred.inputtable' : 'hbase.mapred.datatable',
|
|
86
|
'hbase.mapreduce.inputtable' : 'hbase.mapred.datatable'
|
|
87
|
}
|
|
88
|
</PARAM>
|
|
89
|
</PARAMETERS>
|
|
90
|
<ARCS>
|
|
91
|
<ARC to="finalize" />
|
|
92
|
</ARCS>
|
|
93
|
</NODE>
|
|
94
|
|
|
95
|
<NODE name="finalize" type="FinalizeDedupIndexFeeding">
|
|
96
|
<DESCRIPTION>commit changes</DESCRIPTION>
|
|
97
|
<PARAMETERS />
|
|
98
|
<ARCS>
|
|
99
|
<ARC to="updateDs" />
|
|
100
|
</ARCS>
|
|
101
|
</NODE>
|
|
102
|
|
|
103
|
<NODE name="updateDs" type="IndexDsUpdateJob">
|
|
104
|
<DESCRIPTION>update DS</DESCRIPTION>
|
|
105
|
<PARAMETERS />
|
|
106
|
<ARCS>
|
|
107
|
<ARC to="success" />
|
|
108
|
</ARCS>
|
|
109
|
</NODE>
|
|
110
|
|
|
111
|
</CONFIGURATION>
|
|
112
|
<STATUS />
|
|
113
|
</BODY>
|
|
114
|
</RESOURCE_PROFILE>
|
|
115
|
|
|
116
|
|