Revision 38725
Added by Claudio Atzori about 9 years ago
modules/dnet-deduplication/trunk/src/main/resources/eu/dnetlib/test/profiles/meta/offline.dedup.person.xml | ||
---|---|---|
20 | 20 |
</WORKFLOW> |
21 | 21 |
</WORKFLOW> |
22 | 22 |
<WORKFLOW id="31483043-7dd0-435f-b76e-bad9107aecc4_V29ya2Zsb3dEU1Jlc291cmNlcy9Xb3JrZmxvd0RTUmVzb3VyY2VUeXBl" name="Dedup person full"/> |
23 |
<WORKFLOW id="ae261797-c373-48ce-8581-f1b66f702323_V29ya2Zsb3dEU1Jlc291cmNlcy9Xb3JrZmxvd0RTUmVzb3VyY2VUeXBl" name="Build Anchor Map"/> |
|
24 |
<WORKFLOW id="1c782e6e-33d5-4986-b2f0-5d017415d348_V29ya2Zsb3dEU1Jlc291cmNlcy9Xb3JrZmxvd0RTUmVzb3VyY2VUeXBl" name="Export Fullnames"/> |
|
23 | 25 |
</WORKFLOW> |
24 | 26 |
</CONFIGURATION> |
25 | 27 |
<SCHEDULING enabled="false"> |
modules/dnet-deduplication/trunk/src/main/resources/eu/dnetlib/test/profiles/meta/workflows/copyHbaseTable3.xml | ||
---|---|---|
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="3d12d23f-9f4e-48a4-99c9-d896358a3f7b_V29ya2Zsb3dEU1Jlc291cmNlcy9Xb3JrZmxvd0RTUmVzb3VyY2VUeXBl"/> |
|
5 |
<RESOURCE_TYPE value="WorkflowDSResourceType"/> |
|
6 |
<RESOURCE_KIND value="WorkflowDSResources"/> |
|
7 |
<RESOURCE_URI value=""/> |
|
8 |
<DATE_OF_CREATION value="2006-05-04T18:13:51.0Z"/> |
|
9 |
</HEADER> |
|
10 |
<BODY> |
|
11 |
<WORKFLOW_NAME>Copy HBase Table</WORKFLOW_NAME> |
|
12 |
<WORKFLOW_TYPE>Deduplication</WORKFLOW_TYPE> |
|
13 |
<WORKFLOW_PRIORITY>30</WORKFLOW_PRIORITY> |
|
14 |
<CONFIGURATION start="manual"> |
|
15 |
<NODE name="setInfo" type="PrepareCopyTable" isStart="true"> |
|
16 |
<DESCRIPTION>Set copy parameters</DESCRIPTION> |
|
17 |
<PARAMETERS> |
|
18 |
<PARAM required="true" type="string" name="sourceCluster" managedBy="user" function="obtainValues('hadoopClusters', {})">DM</PARAM> |
|
19 |
<PARAM required="true" type="string" name="sourceTable" managedBy="user"></PARAM> |
|
20 |
<PARAM required="true" type="string" name="targetCluster" managedBy="user" function="obtainValues('hadoopClusters', {})">IIS</PARAM> |
|
21 |
<PARAM required="true" type="string" name="targetTable" managedBy="user"></PARAM> |
|
22 |
</PARAMETERS> |
|
23 |
<ARCS> |
|
24 |
<ARC to="checkSource" /> |
|
25 |
</ARCS> |
|
26 |
</NODE> |
|
27 |
<NODE name="checkSource" type="CheckHBaseTable"> |
|
28 |
<DESCRIPTION>check hbase source table</DESCRIPTION> |
|
29 |
<PARAMETERS> |
|
30 |
<PARAM required="true" type="string" name="tableColumnsParamName" managedBy="system">hTableColumns</PARAM> |
|
31 |
<PARAM required="true" type="string" name="envParams" managedBy="system"> |
|
32 |
{ |
|
33 |
'hbaseTable' : 'sourceTable', |
|
34 |
'cluster' : 'sourceCluster' |
|
35 |
} |
|
36 |
</PARAM> |
|
37 |
<PARAM required="true" type="string" name="existOutNode" managedBy="system">yes</PARAM> |
|
38 |
<PARAM required="true" type="string" name="dontExistOutNode" managedBy="system">no</PARAM> |
|
39 |
</PARAMETERS> |
|
40 |
<ARCS> |
|
41 |
<ARC to="sourceDesc" name="yes" /> |
|
42 |
<ARC to="failure" name ="no" /> |
|
43 |
</ARCS> |
|
44 |
</NODE> |
|
45 |
<NODE name="sourceDesc" type="GetHBaseTableDescription"> |
|
46 |
<DESCRIPTION>get source table description</DESCRIPTION> |
|
47 |
<PARAMETERS> |
|
48 |
<PARAM required="true" type="string" name="tableColumnsParamName" managedBy="system">hTableColumns</PARAM> |
|
49 |
<PARAM required="true" type="string" name="envParams" managedBy="system"> |
|
50 |
{ |
|
51 |
'hbaseTable' : 'sourceTable', |
|
52 |
'cluster' : 'sourceCluster' |
|
53 |
} |
|
54 |
</PARAM> |
|
55 |
</PARAMETERS> |
|
56 |
<ARCS> |
|
57 |
<ARC to="checkTarget" /> |
|
58 |
</ARCS> |
|
59 |
</NODE> |
|
60 |
<NODE name="checkTarget" type="CheckHBaseTable"> |
|
61 |
<DESCRIPTION>check hbase target table</DESCRIPTION> |
|
62 |
<PARAMETERS> |
|
63 |
<PARAM required="true" type="string" name="tableColumnsParamName" managedBy="system">hTableColumns</PARAM> |
|
64 |
<PARAM required="true" type="string" name="envParams" managedBy="system"> |
|
65 |
{ |
|
66 |
'hbaseTable' : 'targetTable', |
|
67 |
'cluster' : 'targetCluster' |
|
68 |
} |
|
69 |
</PARAM> |
|
70 |
<PARAM required="true" type="string" name="existOutNode" managedBy="system">yes</PARAM> |
|
71 |
<PARAM required="true" type="string" name="dontExistOutNode" managedBy="system">no</PARAM> |
|
72 |
</PARAMETERS> |
|
73 |
<ARCS> |
|
74 |
<ARC to="setMode" name="yes" /> |
|
75 |
<ARC to="create" name="no" /> |
|
76 |
</ARCS> |
|
77 |
</NODE> |
|
78 |
<NODE name="setMode" type="Selection"> |
|
79 |
<DESCRIPTION>select copy mode</DESCRIPTION> |
|
80 |
<PARAMETERS> |
|
81 |
<PARAM required="true" type="string" name="selection" managedBy="user" function="validValues(['REFRESH','INCREMENTAL'])">REFRESH</PARAM> |
|
82 |
</PARAMETERS> |
|
83 |
<ARCS> |
|
84 |
<ARC to="drop" name="REFRESH" /> |
|
85 |
<ARC to="copyTable" name="INCREMENTAL" /> |
|
86 |
</ARCS> |
|
87 |
</NODE> |
|
88 |
<NODE name="drop" type="DropHBaseTable"> |
|
89 |
<DESCRIPTION>drop hbase target table</DESCRIPTION> |
|
90 |
<PARAMETERS> |
|
91 |
<PARAM required="true" type="string" name="envParams" managedBy="system"> |
|
92 |
{ |
|
93 |
'hbaseTable' : 'targetTable', |
|
94 |
'cluster' : 'targetCluster' |
|
95 |
} |
|
96 |
</PARAM> |
|
97 |
</PARAMETERS> |
|
98 |
<ARCS> |
|
99 |
<ARC to="create" /> |
|
100 |
</ARCS> |
|
101 |
</NODE> |
|
102 |
<NODE name="create" type="CreateHBaseTable"> |
|
103 |
<DESCRIPTION>create hbase table</DESCRIPTION> |
|
104 |
<PARAMETERS> |
|
105 |
<PARAM name="tableColumnsParamName" type="string" managedBy="system" required="true">hTableColumns</PARAM> |
|
106 |
<PARAM required="true" type="string" name="envParams" managedBy="system"> |
|
107 |
{ |
|
108 |
'hbaseTable' : 'targetTable', |
|
109 |
'cluster' : 'targetCluster' |
|
110 |
} |
|
111 |
</PARAM> |
|
112 |
</PARAMETERS> |
|
113 |
<ARCS> |
|
114 |
<ARC to="copyTable" /> |
|
115 |
</ARCS> |
|
116 |
</NODE> |
|
117 |
<NODE name="copyTable" type="SubmitHadoopJob"> |
|
118 |
<DESCRIPTION>Copy table Job</DESCRIPTION> |
|
119 |
<PARAMETERS> |
|
120 |
<PARAM required="true" type="string" name="hadoopJob" managedBy="system">copytable</PARAM> |
|
121 |
<PARAM required="true" type="string" name="envParams" managedBy="system"> |
|
122 |
{ |
|
123 |
'cluster' : 'sourceCluster', |
|
124 |
'targetCluster' : 'targetCluster', |
|
125 |
'new.name' : 'targetTable', |
|
126 |
'hbase.mapreduce.inputtable' : 'sourceTable', |
|
127 |
'peer.adr' : 'peer.adr' |
|
128 |
} |
|
129 |
</PARAM> |
|
130 |
<PARAM required="true" type="boolean" name="simulation" managedBy="user">false</PARAM> |
|
131 |
</PARAMETERS> |
|
132 |
<ARCS> |
|
133 |
<ARC to="success" /> |
|
134 |
</ARCS> |
|
135 |
</NODE> |
|
136 |
</CONFIGURATION> |
|
137 |
<STATUS /> |
|
138 |
</BODY> |
|
139 |
</RESOURCE_PROFILE> |
modules/dnet-deduplication/trunk/src/main/resources/eu/dnetlib/test/profiles/meta/workflows/build.anchor.map.xml | ||
---|---|---|
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="ae261797-c373-48ce-8581-f1b66f702323_V29ya2Zsb3dEU1Jlc291cmNlcy9Xb3JrZmxvd0RTUmVzb3VyY2VUeXBl"/> |
|
5 |
<RESOURCE_TYPE value="WorkflowDSResourceType"/> |
|
6 |
<RESOURCE_KIND value="WorkflowDSResources"/> |
|
7 |
<RESOURCE_URI value=""/> |
|
8 |
<DATE_OF_CREATION value="2006-05-04T18:13:51.0Z"/> |
|
9 |
</HEADER> |
|
10 |
<BODY> |
|
11 |
<WORKFLOW_NAME>Offline Deduplication</WORKFLOW_NAME> |
|
12 |
<WORKFLOW_TYPE>Deduplication</WORKFLOW_TYPE> |
|
13 |
<WORKFLOW_PRIORITY>30</WORKFLOW_PRIORITY> |
|
14 |
<CONFIGURATION start="manual"> |
|
15 |
|
|
16 |
<NODE name="hadoopConfig" type="SetClusterAndTable" isStart="true"> |
|
17 |
<DESCRIPTION>Set table name</DESCRIPTION> |
|
18 |
<PARAMETERS> |
|
19 |
<PARAM required="true" type="string" name="cluster" managedBy="system">DM</PARAM> |
|
20 |
<PARAM required="true" type="string" name="tableParam" managedBy="system">tableName</PARAM> |
|
21 |
</PARAMETERS> |
|
22 |
<ARCS> |
|
23 |
<ARC to="cleanupMap" /> |
|
24 |
</ARCS> |
|
25 |
</NODE> |
|
26 |
<NODE name="setPath" type="SetHdfsPathJob" isStart="true"> |
|
27 |
<DESCRIPTION>Set map path name</DESCRIPTION> |
|
28 |
<PARAMETERS> |
|
29 |
<PARAM required="true" type="string" name="path" managedBy="user">/tmp/anchorMap</PARAM> |
|
30 |
<PARAM required="true" type="string" name="pathParam" managedBy="system">mapPath</PARAM> |
|
31 |
</PARAMETERS> |
|
32 |
<ARCS> |
|
33 |
<ARC to="cleanupMap" /> |
|
34 |
</ARCS> |
|
35 |
</NODE> |
|
36 |
|
|
37 |
<!-- postprocess --> |
|
38 |
|
|
39 |
<NODE name="cleanupMap" type="DeleteHdfsPathJob" isJoin="true"> |
|
40 |
<DESCRIPTION>hdfs cleanup (xml)</DESCRIPTION> |
|
41 |
<PARAMETERS> |
|
42 |
<PARAM required="true" type="string" name="cluster" managedBy="system">DM</PARAM> |
|
43 |
<PARAM required="true" type="string" name="envParams" managedBy="system"> |
|
44 |
{ |
|
45 |
'path' : 'mapPath' |
|
46 |
} |
|
47 |
</PARAM> |
|
48 |
</PARAMETERS> |
|
49 |
<ARCS> |
|
50 |
<ARC to="buildAnchorMap" /> |
|
51 |
</ARCS> |
|
52 |
</NODE> |
|
53 |
|
|
54 |
<NODE name="buildAnchorMap" type="SubmitHadoopJob"> |
|
55 |
<DESCRIPTION>redirect rels</DESCRIPTION> |
|
56 |
<PARAMETERS> |
|
57 |
<PARAM required="true" type="string" name="hadoopJob" managedBy="system">buildMergedToAnchorMapJob</PARAM> |
|
58 |
<PARAM required="true" type="string" name="envParams" managedBy="system"> |
|
59 |
{ |
|
60 |
'cluster' : 'cluster', |
|
61 |
'hbase.mapred.inputtable' : 'tableName', |
|
62 |
'hbase.mapreduce.inputtable' : 'tableName', |
|
63 |
'hbase.mapred.outputtable' : 'tableName', |
|
64 |
'hbase.mapreduce.outputtable' : 'tableName', |
|
65 |
'mapred.output.dir' : 'mapPath' |
|
66 |
} |
|
67 |
</PARAM> |
|
68 |
</PARAMETERS> |
|
69 |
<ARCS> |
|
70 |
<ARC to="success" /> |
|
71 |
</ARCS> |
|
72 |
</NODE> |
|
73 |
|
|
74 |
|
|
75 |
</CONFIGURATION> |
|
76 |
<STATUS /> |
|
77 |
</BODY> |
|
78 |
</RESOURCE_PROFILE> |
modules/dnet-deduplication/trunk/src/main/resources/eu/dnetlib/test/profiles/meta/workflows/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', |
modules/dnet-deduplication/trunk/src/main/resources/eu/dnetlib/test/profiles/meta/workflows/copyBackGtTable.xml | ||
---|---|---|
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="3d12d23f-9f4e-48a4-99c9-d896358a3f7b_V29ya2Zsb3dEU1Jlc291cmNlcy9Xb3JrZmxvd0RTUmVzb3VyY2VUeXBl"/> |
|
5 |
<RESOURCE_TYPE value="WorkflowDSResourceType"/> |
|
6 |
<RESOURCE_KIND value="WorkflowDSResources"/> |
|
7 |
<RESOURCE_URI value=""/> |
|
8 |
<DATE_OF_CREATION value="2006-05-04T18:13:51.0Z"/> |
|
9 |
</HEADER> |
|
10 |
<BODY> |
|
11 |
<WORKFLOW_NAME>Copy HBase Table</WORKFLOW_NAME> |
|
12 |
<WORKFLOW_TYPE>Deduplication</WORKFLOW_TYPE> |
|
13 |
<WORKFLOW_PRIORITY>30</WORKFLOW_PRIORITY> |
|
14 |
<CONFIGURATION start="manual"> |
|
15 |
|
|
16 |
<NODE name="setDedupConfigs" type="SetDedupConfiguration" isStart="true"> |
|
17 |
<DESCRIPTION>Set Dedup conf</DESCRIPTION> |
|
18 |
<PARAMETERS> |
|
19 |
<PARAM function="obtainValues('dedupOrchestrations', {})" required="true" type="string" name="dedupConfigSequence" managedBy="user"></PARAM> |
|
20 |
<PARAM required="true" type="string" name="dedupConfigSequenceParam" managedBy="system">dedup.conf.queue</PARAM> |
|
21 |
</PARAMETERS> |
|
22 |
<ARCS> |
|
23 |
<ARC to="checkConf" /> |
|
24 |
</ARCS> |
|
25 |
</NODE> |
|
26 |
<NODE name="checkConf" type="DedupCheckConfiguration"> |
|
27 |
<DESCRIPTION/> |
|
28 |
<PARAMETERS> |
|
29 |
<PARAM required="true" type="string" name="dedupConfigSequenceParam" managedBy="system">dedup.conf.queue</PARAM> |
|
30 |
</PARAMETERS> |
|
31 |
<ARCS> |
|
32 |
<ARC to="setInfo" /> |
|
33 |
</ARCS> |
|
34 |
</NODE> |
|
35 |
|
|
36 |
<NODE name="setInfo" type="PrepareCopyTable"> |
|
37 |
<DESCRIPTION>Set copy parameters</DESCRIPTION> |
|
38 |
<PARAMETERS> |
|
39 |
<PARAM required="true" type="string" name="sourceCluster" managedBy="user" function="obtainValues('hadoopClusters', {})">DM</PARAM> |
|
40 |
<PARAM required="true" type="string" name="sourceTable" managedBy="user"></PARAM> |
|
41 |
<PARAM required="true" type="string" name="targetCluster" managedBy="user" function="obtainValues('hadoopClusters', {})">IIS</PARAM> |
|
42 |
<PARAM required="true" type="string" name="targetTable" managedBy="user"></PARAM> |
|
43 |
</PARAMETERS> |
|
44 |
<ARCS> |
|
45 |
<ARC to="checkSource" /> |
|
46 |
</ARCS> |
|
47 |
</NODE> |
|
48 |
<NODE name="checkSource" type="CheckHBaseTable"> |
|
49 |
<DESCRIPTION>check hbase source table</DESCRIPTION> |
|
50 |
<PARAMETERS> |
|
51 |
<PARAM required="true" type="string" name="tableColumnsParamName" managedBy="system">hTableColumns</PARAM> |
|
52 |
<PARAM required="true" type="string" name="envParams" managedBy="system"> |
|
53 |
{ |
|
54 |
'hbaseTable' : 'sourceTable', |
|
55 |
'cluster' : 'sourceCluster' |
|
56 |
} |
|
57 |
</PARAM> |
|
58 |
<PARAM required="true" type="string" name="existOutNode" managedBy="system">yes</PARAM> |
|
59 |
<PARAM required="true" type="string" name="dontExistOutNode" managedBy="system">no</PARAM> |
|
60 |
</PARAMETERS> |
|
61 |
<ARCS> |
|
62 |
<ARC to="sourceDesc" name="yes" /> |
|
63 |
<ARC to="failure" name ="no" /> |
|
64 |
</ARCS> |
|
65 |
</NODE> |
|
66 |
<NODE name="sourceDesc" type="GetHBaseTableDescription"> |
|
67 |
<DESCRIPTION>get source table description</DESCRIPTION> |
|
68 |
<PARAMETERS> |
|
69 |
<PARAM required="true" type="string" name="tableColumnsParamName" managedBy="system">hTableColumns</PARAM> |
|
70 |
<PARAM required="true" type="string" name="envParams" managedBy="system"> |
|
71 |
{ |
|
72 |
'hbaseTable' : 'sourceTable', |
|
73 |
'cluster' : 'sourceCluster' |
|
74 |
} |
|
75 |
</PARAM> |
|
76 |
</PARAMETERS> |
|
77 |
<ARCS> |
|
78 |
<ARC to="checkTarget" /> |
|
79 |
</ARCS> |
|
80 |
</NODE> |
|
81 |
<NODE name="checkTarget" type="CheckHBaseTable"> |
|
82 |
<DESCRIPTION>check hbase target table</DESCRIPTION> |
|
83 |
<PARAMETERS> |
|
84 |
<PARAM required="true" type="string" name="tableColumnsParamName" managedBy="system">hTableColumns</PARAM> |
|
85 |
<PARAM required="true" type="string" name="envParams" managedBy="system"> |
|
86 |
{ |
|
87 |
'hbaseTable' : 'targetTable', |
|
88 |
'cluster' : 'targetCluster' |
|
89 |
} |
|
90 |
</PARAM> |
|
91 |
<PARAM required="true" type="string" name="existOutNode" managedBy="system">yes</PARAM> |
|
92 |
<PARAM required="true" type="string" name="dontExistOutNode" managedBy="system">no</PARAM> |
|
93 |
</PARAMETERS> |
|
94 |
<ARCS> |
|
95 |
<ARC to="setMode" name="yes" /> |
|
96 |
<ARC to="create" name="no" /> |
|
97 |
</ARCS> |
|
98 |
</NODE> |
|
99 |
<NODE name="setMode" type="Selection"> |
|
100 |
<DESCRIPTION>select copy mode</DESCRIPTION> |
|
101 |
<PARAMETERS> |
|
102 |
<PARAM required="true" type="string" name="selection" managedBy="user" function="validValues(['REFRESH','INCREMENTAL'])">REFRESH</PARAM> |
|
103 |
</PARAMETERS> |
|
104 |
<ARCS> |
|
105 |
<ARC to="drop" name="REFRESH" /> |
|
106 |
<ARC to="copyTable" name="INCREMENTAL" /> |
|
107 |
</ARCS> |
|
108 |
</NODE> |
|
109 |
<NODE name="drop" type="DropHBaseTable"> |
|
110 |
<DESCRIPTION>drop hbase target table</DESCRIPTION> |
|
111 |
<PARAMETERS> |
|
112 |
<PARAM required="true" type="string" name="envParams" managedBy="system"> |
|
113 |
{ |
|
114 |
'hbaseTable' : 'targetTable', |
|
115 |
'cluster' : 'targetCluster' |
|
116 |
} |
|
117 |
</PARAM> |
|
118 |
</PARAMETERS> |
|
119 |
<ARCS> |
|
120 |
<ARC to="create" /> |
|
121 |
</ARCS> |
|
122 |
</NODE> |
|
123 |
<NODE name="create" type="CreateHBaseTable"> |
|
124 |
<DESCRIPTION>create hbase table</DESCRIPTION> |
|
125 |
<PARAMETERS> |
|
126 |
<PARAM name="tableColumnsParamName" type="string" managedBy="system" required="true">hTableColumns</PARAM> |
|
127 |
<PARAM required="true" type="string" name="envParams" managedBy="system"> |
|
128 |
{ |
|
129 |
'hbaseTable' : 'targetTable', |
|
130 |
'cluster' : 'targetCluster' |
|
131 |
} |
|
132 |
</PARAM> |
|
133 |
</PARAMETERS> |
|
134 |
<ARCS> |
|
135 |
<ARC to="cleanGT" /> |
|
136 |
</ARCS> |
|
137 |
</NODE> |
|
138 |
|
|
139 |
|
|
140 |
|
|
141 |
<NODE name="cleanGT" type="DedupConfigurationAwareJob"> |
|
142 |
<DESCRIPTION>Clean GT</DESCRIPTION> |
|
143 |
<PARAMETERS> |
|
144 |
<PARAM required="true" type="string" name="hadoopJob" managedBy="system">gtCleanerJob</PARAM> |
|
145 |
<PARAM required="true" type="string" name="dedupConfigSequenceParam" managedBy="system">dedup.conf.queue</PARAM> |
|
146 |
<PARAM required="true" type="string" name="envParams" managedBy="system"> |
|
147 |
{ |
|
148 |
'cluster' : 'sourceCluster', |
|
149 |
'dedup.conf' : 'dedup.conf', |
|
150 |
'entityTypeId' : 'entityTypeId', |
|
151 |
'entityType' : 'entityType', |
|
152 |
'hbase.mapred.inputtable' : 'sourceTable', |
|
153 |
'hbase.mapreduce.inputtable' : 'sourceTable', |
|
154 |
'hbase.mapred.outputtable' : 'sourceTable', |
|
155 |
'hbase.mapreduce.outputtable' : 'sourceTable' |
|
156 |
|
|
157 |
} |
|
158 |
</PARAM> |
|
159 |
<PARAM required="true" type="boolean" name="simulation" managedBy="user">false</PARAM> |
|
160 |
</PARAMETERS> |
|
161 |
<ARCS> |
|
162 |
<ARC to="copyTable" /> |
|
163 |
</ARCS> |
|
164 |
</NODE> |
|
165 |
<NODE name="copyTable" type="SubmitHadoopJob"> |
|
166 |
<DESCRIPTION>Copy table Job</DESCRIPTION> |
|
167 |
<PARAMETERS> |
|
168 |
<PARAM required="true" type="string" name="hadoopJob" managedBy="system">copytable</PARAM> |
|
169 |
<PARAM required="true" type="string" name="envParams" managedBy="system"> |
|
170 |
{ |
|
171 |
'cluster' : 'sourceCluster', |
|
172 |
'targetCluster' : 'targetCluster', |
|
173 |
'new.name' : 'targetTable', |
|
174 |
'hbase.mapreduce.inputtable' : 'sourceTable', |
|
175 |
'peer.adr' : 'peer.adr' |
|
176 |
} |
|
177 |
</PARAM> |
|
178 |
<PARAM required="true" type="boolean" name="simulation" managedBy="user">false</PARAM> |
|
179 |
</PARAMETERS> |
|
180 |
<ARCS> |
|
181 |
<ARC to="success" /> |
|
182 |
</ARCS> |
|
183 |
</NODE> |
|
184 |
</CONFIGURATION> |
|
185 |
<STATUS /> |
|
186 |
</BODY> |
|
187 |
</RESOURCE_PROFILE> |
modules/dnet-deduplication/trunk/src/main/resources/eu/dnetlib/test/profiles/meta/workflows/export.fullnames.xml | ||
---|---|---|
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="1c782e6e-33d5-4986-b2f0-5d017415d348_V29ya2Zsb3dEU1Jlc291cmNlcy9Xb3JrZmxvd0RTUmVzb3VyY2VUeXBl"/> |
|
5 |
<RESOURCE_TYPE value="WorkflowDSResourceType"/> |
|
6 |
<RESOURCE_KIND value="WorkflowDSResources"/> |
|
7 |
<RESOURCE_URI value=""/> |
|
8 |
<DATE_OF_CREATION value="2006-05-04T18:13:51.0Z"/> |
|
9 |
</HEADER> |
|
10 |
<BODY> |
|
11 |
<WORKFLOW_NAME>Export Fullnames</WORKFLOW_NAME> |
|
12 |
<WORKFLOW_TYPE>Deduplication</WORKFLOW_TYPE> |
|
13 |
<WORKFLOW_PRIORITY>30</WORKFLOW_PRIORITY> |
|
14 |
<CONFIGURATION start="manual"> |
|
15 |
|
|
16 |
<NODE name="setDedupConfigs" type="SetDedupConfiguration" isStart="true"> |
|
17 |
<DESCRIPTION>Set Dedup conf</DESCRIPTION> |
|
18 |
<PARAMETERS> |
|
19 |
<PARAM function="obtainValues('dedupOrchestrations', {})" required="true" type="string" name="dedupConfigSequence" managedBy="user"></PARAM> |
|
20 |
<PARAM required="true" type="string" name="dedupConfigSequenceParam" managedBy="system">dedup.conf.queue</PARAM> |
|
21 |
</PARAMETERS> |
|
22 |
<ARCS> |
|
23 |
<ARC to="checkConf" /> |
|
24 |
</ARCS> |
|
25 |
</NODE> |
|
26 |
|
|
27 |
<NODE name="hadoopConfig" type="SetClusterAndTable" isStart="true"> |
|
28 |
<DESCRIPTION>Set table name</DESCRIPTION> |
|
29 |
<PARAMETERS> |
|
30 |
<PARAM required="true" type="string" name="cluster" managedBy="system">DM</PARAM> |
|
31 |
<PARAM required="true" type="string" name="tableParam" managedBy="system">tableName</PARAM> |
|
32 |
<PARAM required="true" type="string" name="table" managedBy="user"></PARAM> |
|
33 |
</PARAMETERS> |
|
34 |
<ARCS> |
|
35 |
<ARC to="checkConf" /> |
|
36 |
</ARCS> |
|
37 |
</NODE> |
|
38 |
<NODE name="setPath" type="SetHdfsPathJob" isStart="true"> |
|
39 |
<DESCRIPTION>Set map path name</DESCRIPTION> |
|
40 |
<PARAMETERS> |
|
41 |
<PARAM required="true" type="string" name="path" managedBy="user">/tmp/fullnames_test</PARAM> |
|
42 |
<PARAM required="true" type="string" name="pathParam" managedBy="system">exportPath</PARAM> |
|
43 |
</PARAMETERS> |
|
44 |
<ARCS> |
|
45 |
<ARC to="checkConf" /> |
|
46 |
</ARCS> |
|
47 |
</NODE> |
|
48 |
|
|
49 |
<NODE name="checkConf" type="DedupCheckConfiguration" isJoin="true"> |
|
50 |
<DESCRIPTION/> |
|
51 |
<PARAMETERS> |
|
52 |
<PARAM required="true" type="string" name="dedupConfigSequenceParam" managedBy="system">dedup.conf.queue</PARAM> |
|
53 |
</PARAMETERS> |
|
54 |
<ARCS> |
|
55 |
<ARC to="cleanupNames" /> |
|
56 |
</ARCS> |
|
57 |
</NODE> |
|
58 |
|
|
59 |
<NODE name="cleanupNames" type="DeleteHdfsPathJob"> |
|
60 |
<DESCRIPTION>hdfs cleanup</DESCRIPTION> |
|
61 |
<PARAMETERS> |
|
62 |
<PARAM required="true" type="string" name="cluster" managedBy="system">DM</PARAM> |
|
63 |
<PARAM required="true" type="string" name="envParams" managedBy="system"> |
|
64 |
{ |
|
65 |
'path' : 'exportPath' |
|
66 |
} |
|
67 |
</PARAM> |
|
68 |
</PARAMETERS> |
|
69 |
<ARCS> |
|
70 |
<ARC to="exportFullnames" /> |
|
71 |
</ARCS> |
|
72 |
</NODE> |
|
73 |
|
|
74 |
<NODE name="exportFullnames" type="DedupConfigurationAwareJob"> |
|
75 |
<DESCRIPTION>redirect rels</DESCRIPTION> |
|
76 |
<PARAMETERS> |
|
77 |
<PARAM required="true" type="string" name="hadoopJob" managedBy="system">exportPersonFullnamesJob</PARAM> |
|
78 |
<PARAM required="true" type="string" name="dedupConfigSequenceParam" managedBy="system">dedup.conf.queue</PARAM> |
|
79 |
<PARAM required="true" type="string" name="envParams" managedBy="system"> |
|
80 |
{ |
|
81 |
'cluster' : 'cluster', |
|
82 |
'entityTypeId' : 'entityTypeId', |
|
83 |
'entityType' : 'entityType', |
|
84 |
'hbase.mapred.inputtable' : 'tableName', |
|
85 |
'hbase.mapreduce.inputtable' : 'tableName', |
|
86 |
'hbase.mapred.outputtable' : 'tableName', |
|
87 |
'hbase.mapreduce.outputtable' : 'tableName', |
|
88 |
'mapred.output.dir' : 'exportPath' |
|
89 |
} |
|
90 |
</PARAM> |
|
91 |
</PARAMETERS> |
|
92 |
<ARCS> |
|
93 |
<ARC to="success" /> |
|
94 |
</ARCS> |
|
95 |
</NODE> |
|
96 |
|
|
97 |
|
|
98 |
</CONFIGURATION> |
|
99 |
<STATUS /> |
|
100 |
</BODY> |
|
101 |
</RESOURCE_PROFILE> |
modules/dnet-deduplication/trunk/src/main/resources/eu/dnetlib/test/profiles/meta/workflows/delete.simrels.xml | ||
---|---|---|
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="0fc975d7-65c1-4d39-88ec-7bca72744e89_V29ya2Zsb3dEU1Jlc291cmNlcy9Xb3JrZmxvd0RTUmVzb3VyY2VUeXBl"/> |
|
5 |
<RESOURCE_TYPE value="WorkflowDSResourceType"/> |
|
6 |
<RESOURCE_KIND value="WorkflowDSResources"/> |
|
7 |
<RESOURCE_URI value=""/> |
|
8 |
<DATE_OF_CREATION value="2006-05-04T18:13:51.0Z"/> |
|
9 |
</HEADER> |
|
10 |
<BODY> |
|
11 |
<WORKFLOW_NAME>Delete Similarity Rels</WORKFLOW_NAME> |
|
12 |
<WORKFLOW_TYPE>Deduplication</WORKFLOW_TYPE> |
|
13 |
<WORKFLOW_PRIORITY>30</WORKFLOW_PRIORITY> |
|
14 |
<CONFIGURATION start="manual"> |
|
15 |
|
|
16 |
<NODE name="setDedupConfigs" type="SetDedupConfiguration" isStart="true"> |
|
17 |
<DESCRIPTION>Set Dedup conf</DESCRIPTION> |
|
18 |
<PARAMETERS> |
|
19 |
<PARAM function="obtainValues('dedupOrchestrations', {})" required="true" type="string" name="dedupConfigSequence" managedBy="user"></PARAM> |
|
20 |
<PARAM required="true" type="string" name="dedupConfigSequenceParam" managedBy="system">dedup.conf.queue</PARAM> |
|
21 |
</PARAMETERS> |
|
22 |
<ARCS> |
|
23 |
<ARC to="checkConf" /> |
|
24 |
</ARCS> |
|
25 |
</NODE> |
|
26 |
<NODE name="hadoopConfig" type="SetClusterAndTable" isStart="true"> |
|
27 |
<DESCRIPTION>Set table name</DESCRIPTION> |
|
28 |
<PARAMETERS> |
|
29 |
<PARAM required="true" type="string" name="cluster" managedBy="system">DM</PARAM> |
|
30 |
<PARAM required="true" type="string" name="tableParam" managedBy="system">tableName</PARAM> |
|
31 |
</PARAMETERS> |
|
32 |
<ARCS> |
|
33 |
<ARC to="checkConf" /> |
|
34 |
</ARCS> |
|
35 |
</NODE> |
|
36 |
|
|
37 |
<NODE name="checkConf" type="DedupCheckConfiguration" isJoin="true"> |
|
38 |
<DESCRIPTION/> |
|
39 |
<PARAMETERS> |
|
40 |
<PARAM required="true" type="string" name="dedupConfigSequenceParam" managedBy="system">dedup.conf.queue</PARAM> |
|
41 |
</PARAMETERS> |
|
42 |
<ARCS> |
|
43 |
<ARC to="deleteSimRels" /> |
|
44 |
</ARCS> |
|
45 |
</NODE> |
|
46 |
|
|
47 |
<NODE name="deleteSimRels" type="DedupConfigurationAwareJob"> |
|
48 |
<DESCRIPTION>redirect rels</DESCRIPTION> |
|
49 |
<PARAMETERS> |
|
50 |
<PARAM required="true" type="string" name="hadoopJob" managedBy="system">deleteSimRelJob</PARAM> |
|
51 |
<PARAM required="true" type="string" name="dedupConfigSequenceParam" managedBy="system">dedup.conf.queue</PARAM> |
|
52 |
<PARAM required="true" type="string" name="envParams" managedBy="system"> |
|
53 |
{ |
|
54 |
'dedup.conf' : 'dedup.conf', |
|
55 |
'entityTypeId' : 'entityTypeId', |
|
56 |
'entityType' : 'entityType', |
|
57 |
'cluster' : 'cluster', |
|
58 |
'hbase.mapred.inputtable' : 'tableName', |
|
59 |
'hbase.mapreduce.inputtable' : 'tableName', |
|
60 |
'hbase.mapred.outputtable' : 'tableName', |
|
61 |
'hbase.mapreduce.outputtable' : 'tableName' |
|
62 |
} |
|
63 |
</PARAM> |
|
64 |
</PARAMETERS> |
|
65 |
<ARCS> |
|
66 |
<ARC to="success" /> |
|
67 |
</ARCS> |
|
68 |
</NODE> |
|
69 |
|
|
70 |
|
|
71 |
</CONFIGURATION> |
|
72 |
<STATUS /> |
|
73 |
</BODY> |
|
74 |
</RESOURCE_PROFILE> |
Also available in: Unified diff
added dedup person workflows