1
|
<?xml version="1.0" encoding="UTF-8"?>
|
2
|
<RESOURCE_PROFILE>
|
3
|
<HEADER>
|
4
|
<RESOURCE_IDENTIFIER
|
5
|
value="c7d7d775-2db3-474d-85ab-5173a582d515_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
|
</HEADER>
|
11
|
<BODY>
|
12
|
<WORKFLOW_NAME>OAF to HBase</WORKFLOW_NAME>
|
13
|
<WORKFLOW_TYPE>Data Load</WORKFLOW_TYPE>
|
14
|
<WORKFLOW_PRIORITY>30</WORKFLOW_PRIORITY>
|
15
|
<CONFIGURATION start="manual">
|
16
|
<NODE name="prepareImport" type="PrepareMDStoreImport" isStart="true">
|
17
|
<DESCRIPTION>Configure export to HDFS</DESCRIPTION>
|
18
|
<PARAMETERS>
|
19
|
<PARAM required="true" type="string" name="hdfsPathParam" managedBy="system">hdfsPath</PARAM>
|
20
|
<PARAM required="true" type="string" name="hdfsPath" managedBy="system">/tmp/mdstores_oaf-store-cleaned.seq</PARAM>
|
21
|
<PARAM required="true" type="string" name="xsltParam" managedBy="system">xslt</PARAM>
|
22
|
<PARAM required="true" type="string" name="xslt" managedBy="system">/eu/dnetlib/data/transform/dmf_2_hbase.xsl</PARAM>
|
23
|
</PARAMETERS>
|
24
|
<ARCS>
|
25
|
<ARC to="reuseHdfsRecords"/>
|
26
|
</ARCS>
|
27
|
</NODE>
|
28
|
<NODE name="reuseHdfsRecords" type="ReuseHdfsRecords">
|
29
|
<DESCRIPTION>reuse mdstore records</DESCRIPTION>
|
30
|
<PARAMETERS>
|
31
|
<PARAM required="true" type="boolean" name="reuseMdRecords" managedBy="user">false</PARAM>
|
32
|
</PARAMETERS>
|
33
|
<ARCS>
|
34
|
<ARC name="true" to="mapreduce"/>
|
35
|
<ARC name="false" to="exportRecords"/>
|
36
|
</ARCS>
|
37
|
</NODE>
|
38
|
<NODE name="exportRecords" type="MDStoreBatchExporter">
|
39
|
<DESCRIPTION>Fetch mdstore records</DESCRIPTION>
|
40
|
<PARAMETERS>
|
41
|
<PARAM required="true" type="string" name="format" managedBy="system">OAF</PARAM>
|
42
|
<PARAM required="true" type="string" name="layout" managedBy="system">store</PARAM>
|
43
|
<PARAM required="true" type="string" name="interpretation" managedBy="system">cleaned</PARAM>
|
44
|
<PARAM required="true" type="string" name="outputEprParam" managedBy="system">records_epr</PARAM>
|
45
|
</PARAMETERS>
|
46
|
<ARCS>
|
47
|
<ARC to="storeHdfsRecords"/>
|
48
|
</ARCS>
|
49
|
</NODE>
|
50
|
<NODE name="storeHdfsRecords" type="StoreHdfsRecords">
|
51
|
<DESCRIPTION>Store records to HDFS</DESCRIPTION>
|
52
|
<PARAMETERS>
|
53
|
<PARAM required="true" type="string" name="inputEprParam" managedBy="system">records_epr</PARAM>
|
54
|
<PARAM required="true" type="string" name="hdfsPathParam" managedBy="system">hdfsPath</PARAM>
|
55
|
<PARAM required="true" type="string" name="cluster" managedBy="system">DM</PARAM>
|
56
|
</PARAMETERS>
|
57
|
<ARCS>
|
58
|
<ARC to="mapreduce"/>
|
59
|
</ARCS>
|
60
|
</NODE>
|
61
|
<NODE name="mapreduce" type="SubmitHadoopJob">
|
62
|
<DESCRIPTION>Run M/R import Job</DESCRIPTION>
|
63
|
<PARAMETERS>
|
64
|
<PARAM required="true" type="string" name="cluster" managedBy="system">DM</PARAM>
|
65
|
<PARAM required="true" type="string" name="hadoopJob" managedBy="system">mdStoreHdfsImportJob</PARAM>
|
66
|
<PARAM required="true" type="string" name="envParams" managedBy="system">
|
67
|
{
|
68
|
'mapred.input.dir' : 'hdfsPath',
|
69
|
'hbase.import.xslt' : 'xslt'
|
70
|
}
|
71
|
</PARAM>
|
72
|
<PARAM required="true" type="string" name="sysParams" managedBy="system">
|
73
|
{
|
74
|
'hbase.mapred.outputtable' : 'hbase.mapred.datatable'
|
75
|
}
|
76
|
</PARAM>
|
77
|
<PARAM required="true" type="boolean" name="simulation" managedBy="user">false</PARAM>
|
78
|
</PARAMETERS>
|
79
|
<ARCS>
|
80
|
<ARC to="checkOAFResultCount"/>
|
81
|
</ARCS>
|
82
|
</NODE>
|
83
|
<NODE name="checkOAFResultCount" type="CheckHDFSCount">
|
84
|
<DESCRIPTION>Checks if the number of publications stored on HBASE is the same as those in the HDFS file</DESCRIPTION>
|
85
|
<PARAMETERS>
|
86
|
<PARAM required="true" type="string" name="numberToVerifyParamName" managedBy="system">mdstore.result</PARAM>
|
87
|
<PARAM required="false" type="string" name="hdfsCounterParamName" managedBy="system">mainlog:storeHdfsRecords:count</PARAM>
|
88
|
</PARAMETERS>
|
89
|
<ARCS>
|
90
|
<ARC to="success"/>
|
91
|
</ARCS>
|
92
|
</NODE>
|
93
|
</CONFIGURATION>
|
94
|
<STATUS/>
|
95
|
</BODY>
|
96
|
</RESOURCE_PROFILE>
|
97
|
|
98
|
|