Project

General

Profile

« Previous | Next » 

Revision 38725

added dedup person workflows

View differences:

dedup.roots.export.xml
9 9
	</HEADER>
10 10
	<BODY>
11 11
		<WORKFLOW_NAME>Deduplication export</WORKFLOW_NAME>
12
		<WORKFLOW_TYPE>Deduplication</WORKFLOW_TYPE>
12
		<WORKFLOW_TYPE>Export to HDFS</WORKFLOW_TYPE>
13 13
		<WORKFLOW_PRIORITY>30</WORKFLOW_PRIORITY>
14 14
		<CONFIGURATION start="manual">
15 15
        	<NODE name="setCsvPath" type="SetEnvParameter" isStart="true">
......
39 39
				<PARAMETERS>
40 40
					<PARAM required="true" type="string" name="cluster" managedBy="system">DM</PARAM>
41 41
					<PARAM required="true" type="string" name="tableParam" managedBy="system">tableName</PARAM>
42
					<PARAM required="true" type="string" name="table" managedBy="user"/>
42 43
				</PARAMETERS>
43 44
				<ARCS>
44 45
					<ARC to="cleanupCsv" />
......
59 60
					<ARC to="export" />
60 61
				</ARCS>
61 62
			</NODE>
62
				
63
			
63 64
			<NODE name="export" type="SubmitHadoopJob">
64 65
				<DESCRIPTION>export the representative publications</DESCRIPTION>
65 66
				<PARAMETERS>
66 67
					<PARAM required="true" type="string" name="hadoopJob" managedBy="system">dedupRootsExportJob</PARAM>
67
					<PARAM required="true" type="string" name="sysParams" managedBy="system">
68
						{ 	
69
							'hbase.mapred.inputtable' : 'hbase.mapred.datatable', 
70
							'hbase.mapreduce.inputtable' : 'hbase.mapred.datatable'
71
						}
72
					</PARAM>					
73 68
					<PARAM required="true" type="string" name="envParams" managedBy="system">
74 69
						{ 	
75 70
							'cluster' : 'cluster',

Also available in: Unified diff