1 |
31072
|
claudio.at
|
<?xml version="1.0" encoding="UTF-8"?>
|
2 |
48139
|
alessia.ba
|
<RESOURCE_PROFILE>
|
3 |
31072
|
claudio.at
|
<HEADER>
|
4 |
|
|
<RESOURCE_IDENTIFIER
|
5 |
48139
|
alessia.ba
|
value="939f1075-70ed-4fbd-9055-ea16e5984531_V29ya2Zsb3dEU1Jlc291cmNlcy9Xb3JrZmxvd0RTUmVzb3VyY2VUeXBl"/>
|
6 |
|
|
<RESOURCE_TYPE value="WorkflowDSResourceType"/>
|
7 |
|
|
<RESOURCE_KIND value="WorkflowDSResources"/>
|
8 |
|
|
<RESOURCE_URI value=""/>
|
9 |
|
|
<DATE_OF_CREATION value="2006-05-04T18:13:51.0Z"/>
|
10 |
31072
|
claudio.at
|
</HEADER>
|
11 |
|
|
<BODY>
|
12 |
|
|
<WORKFLOW_NAME>ODF to HBase</WORKFLOW_NAME>
|
13 |
|
|
<WORKFLOW_TYPE>Data Load</WORKFLOW_TYPE>
|
14 |
|
|
<WORKFLOW_PRIORITY>30</WORKFLOW_PRIORITY>
|
15 |
|
|
<CONFIGURATION start="manual">
|
16 |
40430
|
claudio.at
|
<NODE name="setTable" type="SetHBaseTable" isStart="true">
|
17 |
|
|
<DESCRIPTION>set hbase table</DESCRIPTION>
|
18 |
|
|
<PARAMETERS>
|
19 |
|
|
<PARAM name="table" type="string" managedBy="user" required="true"></PARAM>
|
20 |
|
|
</PARAMETERS>
|
21 |
|
|
<ARCS>
|
22 |
|
|
<ARC to="mapreduce"/>
|
23 |
|
|
</ARCS>
|
24 |
|
|
</NODE>
|
25 |
|
|
|
26 |
31072
|
claudio.at
|
<NODE name="prepareImport" type="PrepareMDStoreImport" isStart="true">
|
27 |
|
|
<DESCRIPTION>Configure export to HDFS</DESCRIPTION>
|
28 |
|
|
<PARAMETERS>
|
29 |
|
|
<PARAM required="true" type="string" name="hdfsPathParam" managedBy="system">hdfsPath</PARAM>
|
30 |
40430
|
claudio.at
|
<PARAM required="true" type="string" name="hdfsPath" managedBy="user">/tmp/mdstores_odf-store-cleaned.seq</PARAM>
|
31 |
|
|
<PARAM required="true" type="string" name="mappingParam" managedBy="system">xslt</PARAM>
|
32 |
|
|
<PARAM required="true" type="string" name="mapping" managedBy="user" function="obtainValues('odf2hbaseMappings', {})"></PARAM>
|
33 |
31072
|
claudio.at
|
</PARAMETERS>
|
34 |
|
|
<ARCS>
|
35 |
48139
|
alessia.ba
|
<ARC to="reuseHdfsRecords"/>
|
36 |
31072
|
claudio.at
|
</ARCS>
|
37 |
|
|
</NODE>
|
38 |
|
|
<NODE name="reuseHdfsRecords" type="ReuseHdfsRecords">
|
39 |
|
|
<DESCRIPTION>reuse mdstore records</DESCRIPTION>
|
40 |
|
|
<PARAMETERS>
|
41 |
|
|
<PARAM required="true" type="boolean" name="reuseMdRecords" managedBy="user">false</PARAM>
|
42 |
|
|
</PARAMETERS>
|
43 |
|
|
<ARCS>
|
44 |
40430
|
claudio.at
|
<ARC name="true" to="doneExport"/>
|
45 |
48139
|
alessia.ba
|
<ARC name="false" to="exportRecords"/>
|
46 |
31072
|
claudio.at
|
</ARCS>
|
47 |
|
|
</NODE>
|
48 |
|
|
<NODE name="exportRecords" type="MDStoreBatchExporter">
|
49 |
|
|
<DESCRIPTION>Fetch mdstore records</DESCRIPTION>
|
50 |
|
|
<PARAMETERS>
|
51 |
|
|
<PARAM required="true" type="string" name="format" managedBy="system">ODF</PARAM>
|
52 |
|
|
<PARAM required="true" type="string" name="layout" managedBy="system">store</PARAM>
|
53 |
|
|
<PARAM required="true" type="string" name="interpretation" managedBy="system">cleaned</PARAM>
|
54 |
|
|
<PARAM required="true" type="string" name="outputEprParam" managedBy="system">records_epr</PARAM>
|
55 |
|
|
</PARAMETERS>
|
56 |
|
|
<ARCS>
|
57 |
48139
|
alessia.ba
|
<ARC to="storeHdfsRecords"/>
|
58 |
31072
|
claudio.at
|
</ARCS>
|
59 |
|
|
</NODE>
|
60 |
|
|
<NODE name="storeHdfsRecords" type="StoreHdfsRecords">
|
61 |
|
|
<DESCRIPTION>Store records to HDFS</DESCRIPTION>
|
62 |
|
|
<PARAMETERS>
|
63 |
|
|
<PARAM required="true" type="string" name="inputEprParam" managedBy="system">records_epr</PARAM>
|
64 |
|
|
<PARAM required="true" type="string" name="hdfsPathParam" managedBy="system">hdfsPath</PARAM>
|
65 |
48139
|
alessia.ba
|
<PARAM required="true" type="string" name="cluster" managedBy="system">DM</PARAM>
|
66 |
31072
|
claudio.at
|
</PARAMETERS>
|
67 |
|
|
<ARCS>
|
68 |
40430
|
claudio.at
|
<ARC to="doneExport"/>
|
69 |
|
|
</ARCS>
|
70 |
|
|
</NODE>
|
71 |
|
|
<NODE name="doneExport">
|
72 |
|
|
<DESCRIPTION></DESCRIPTION>
|
73 |
|
|
<PARAMETERS/>
|
74 |
|
|
<ARCS>
|
75 |
48139
|
alessia.ba
|
<ARC to="mapreduce"/>
|
76 |
31072
|
claudio.at
|
</ARCS>
|
77 |
40430
|
claudio.at
|
</NODE>
|
78 |
|
|
<NODE name="mapreduce" type="SubmitHadoopJob" isJoin="true">
|
79 |
31072
|
claudio.at
|
<DESCRIPTION>Run M/R import Job</DESCRIPTION>
|
80 |
|
|
<PARAMETERS>
|
81 |
|
|
<PARAM required="true" type="string" name="cluster" managedBy="system">DM</PARAM>
|
82 |
|
|
<PARAM required="true" type="string" name="hadoopJob" managedBy="system">mdStoreHdfsImportJob</PARAM>
|
83 |
40430
|
claudio.at
|
<PARAM required="true" type="boolean" name="simulation" managedBy="user">false</PARAM>
|
84 |
31072
|
claudio.at
|
<PARAM required="true" type="string" name="envParams" managedBy="system">
|
85 |
40430
|
claudio.at
|
{
|
86 |
|
|
'mapred.input.dir' : 'hdfsPath',
|
87 |
|
|
'hbase.import.xslt' : 'xslt',
|
88 |
|
|
'hbase.mapred.outputtable' : 'hbaseTable'
|
89 |
31072
|
claudio.at
|
}
|
90 |
|
|
</PARAM>
|
91 |
|
|
</PARAMETERS>
|
92 |
|
|
<ARCS>
|
93 |
48139
|
alessia.ba
|
<ARC to="success"/>
|
94 |
35830
|
alessia.ba
|
</ARCS>
|
95 |
35829
|
alessia.ba
|
</NODE>
|
96 |
31072
|
claudio.at
|
</CONFIGURATION>
|
97 |
48139
|
alessia.ba
|
<STATUS/>
|
98 |
31072
|
claudio.at
|
</BODY>
|
99 |
|
|
</RESOURCE_PROFILE>
|
100 |
|
|
|