|
1
|
<RESOURCE_PROFILE>
|
|
2
|
<HEADER>
|
|
3
|
<RESOURCE_IDENTIFIER value="5b05a65a-4eeb-4862-bc55-b35c7ec3baf0_V29ya2Zsb3dUZW1wbGF0ZURTUmVzb3VyY2VzL1dvcmtmbG93VGVtcGxhdGVEU1Jlc291cmNlVHlwZQ=="/>
|
|
4
|
<RESOURCE_TYPE value="WorkflowTemplateDSResourceType"/>
|
|
5
|
<RESOURCE_KIND value="WorkflowTemplateDSResources"/>
|
|
6
|
<RESOURCE_URI value=""/>
|
|
7
|
<DATE_OF_CREATION value="2016-06-20T07:52:08+00:00"/>
|
|
8
|
</HEADER>
|
|
9
|
<BODY>
|
|
10
|
<CONFIGURATION>
|
|
11
|
<PARAMETERS>
|
|
12
|
<PARAM name="reuseMdRecords" description="reuse metadata records?" required="true" type="boolean"/>
|
|
13
|
<PARAM name="mdFormat" description="metadata format name" required="true" type="string"/>
|
|
14
|
<PARAM name="interpretation" description="metadata interpretation name" required="false" type="string" default="cleaned"/>
|
|
15
|
<PARAM name="mapping" description="xslt mapping function" required="true" type="string"/>
|
|
16
|
<PARAM name="hdfsPath" description="HDFS path" required="true" type="string"/>
|
|
17
|
<PARAM name="hbaseTable" description="HBase table name" required="true" type="string"/>
|
|
18
|
<PARAM name="cluster" description="Hadoop cluster name" required="true" type="string"/>
|
|
19
|
</PARAMETERS>
|
|
20
|
<WORKFLOW>
|
|
21
|
|
|
22
|
<NODE name="reuseHdfsRecords" type="ReuseHdfsRecords" isStart="true">
|
|
23
|
<DESCRIPTION>reuse mdstore records</DESCRIPTION>
|
|
24
|
<PARAMETERS>
|
|
25
|
<PARAM name="reuseMdRecords" ref="reuseMdRecords"/>
|
|
26
|
</PARAMETERS>
|
|
27
|
<ARCS>
|
|
28
|
<ARC name="true" to="doneExport"/>
|
|
29
|
<ARC name="false" to="exportRecords"/>
|
|
30
|
</ARCS>
|
|
31
|
</NODE>
|
|
32
|
<NODE name="exportRecords" type="MDStoreBatchExporter">
|
|
33
|
<DESCRIPTION>Fetch mdstore records</DESCRIPTION>
|
|
34
|
<PARAMETERS>
|
|
35
|
<PARAM name="format" ref="mdFormat"/>
|
|
36
|
<PARAM name="layout" value="store"/>
|
|
37
|
<PARAM name="interpretation" ref="interpretation"/>
|
|
38
|
<PARAM name="outputEprParam" value="records_epr"/>
|
|
39
|
</PARAMETERS>
|
|
40
|
<ARCS>
|
|
41
|
<ARC to="storeHdfsRecords"/>
|
|
42
|
</ARCS>
|
|
43
|
</NODE>
|
|
44
|
<NODE name="storeHdfsRecords" type="StoreHdfsRecords">
|
|
45
|
<DESCRIPTION>Store records to HDFS</DESCRIPTION>
|
|
46
|
<PARAMETERS>
|
|
47
|
<PARAM name="inputEprParam" value="records_epr"/>
|
|
48
|
<PARAM name="hdfsPath" ref="hdfsPath"/>
|
|
49
|
<PARAM name="cluster" ref="cluster"/>
|
|
50
|
</PARAMETERS>
|
|
51
|
<ARCS>
|
|
52
|
<ARC to="doneExport"/>
|
|
53
|
</ARCS>
|
|
54
|
</NODE>
|
|
55
|
<NODE name="doneExport">
|
|
56
|
<DESCRIPTION/>
|
|
57
|
<PARAMETERS/>
|
|
58
|
<ARCS>
|
|
59
|
<ARC to="loadMapping"/>
|
|
60
|
</ARCS>
|
|
61
|
</NODE>
|
|
62
|
|
|
63
|
<NODE name="loadMapping" type="LoadMappingProfile">
|
|
64
|
<DESCRIPTION>Load mapping profile</DESCRIPTION>
|
|
65
|
<PARAMETERS>
|
|
66
|
<PARAM name="profileId" ref="mapping"/>
|
|
67
|
<PARAM name="loadedMappingParam" value="hbase.import.xslt"/>
|
|
68
|
</PARAMETERS>
|
|
69
|
<ARCS>
|
|
70
|
<ARC to="mapreduce"/>
|
|
71
|
</ARCS>
|
|
72
|
</NODE>
|
|
73
|
|
|
74
|
<NODE name="mapreduce" type="SubmitHadoopJob">
|
|
75
|
<DESCRIPTION>Run M/R import Job</DESCRIPTION>
|
|
76
|
<PARAMETERS>
|
|
77
|
<PARAM name="cluster" ref="cluster"/>
|
|
78
|
<PARAM name="hadoopJob" value="mdStoreHdfsImportJob"/>
|
|
79
|
<PARAM name="simulation" value="false"/>
|
|
80
|
<PARAM name="jobParams">
|
|
81
|
<MAP>
|
|
82
|
<ENTRY key="mapred.input.dir" ref="hdfsPath"/>
|
|
83
|
<ENTRY key="hbase.mapred.outputtable" ref="hbaseTable"/>
|
|
84
|
<ENTRY key="hbase.import.xslt" env="hbase.import.xslt"/>
|
|
85
|
</MAP>
|
|
86
|
</PARAM>
|
|
87
|
</PARAMETERS>
|
|
88
|
<ARCS>
|
|
89
|
<ARC to="success"/>
|
|
90
|
</ARCS>
|
|
91
|
</NODE>
|
|
92
|
|
|
93
|
</WORKFLOW>
|
|
94
|
</CONFIGURATION>
|
|
95
|
</BODY>
|
|
96
|
</RESOURCE_PROFILE>
|