1
|
<NODE name="SET_INFO" isStart="true" type="SetProviderInfo">
|
2
|
<DESCRIPTION>Set information about current provider</DESCRIPTION>
|
3
|
<PARAMETERS>
|
4
|
<PARAM required="true" type="string" name="providerId" managedBy="system">$params.("dataprovider:id")$</PARAM>
|
5
|
<PARAM required="true" type="string" name="providerName" managedBy="system">$params.("dataprovider:name")$</PARAM>
|
6
|
<PARAM required="true" type="string" name="api" managedBy="system">$params.("dataprovider:interface")$</PARAM>
|
7
|
</PARAMETERS>
|
8
|
<ARCS>
|
9
|
<ARC to="obtainParams"/>
|
10
|
</ARCS>
|
11
|
</NODE>
|
12
|
|
13
|
<NODE name="obtainParams" type="ObtainOpenaireDataSourceParams">
|
14
|
<DESCRIPTION>Obtain data source params</DESCRIPTION>
|
15
|
<PARAMETERS>
|
16
|
<PARAM required="true" type="string" name="providerId" managedBy="system">$params.("dataprovider:id")$</PARAM>
|
17
|
</PARAMETERS>
|
18
|
<ARCS>
|
19
|
<ARC to="PREPARE_ENV_COLLECTION"/>
|
20
|
</ARCS>
|
21
|
</NODE>
|
22
|
|
23
|
<NODE name="PREPARE_ENV_COLLECTION" type="PrepareEnvCollectHadoopJobNode">
|
24
|
<DESCRIPTION>Set in the environment all the variable needed to the collection oozie job</DESCRIPTION>
|
25
|
<PARAMETERS>
|
26
|
<PARAM required="true" type="string" name="mdId" managedBy="system" category="MDSTORE_ID">$params.("harv_id")$</PARAM>
|
27
|
<PARAM required="true" type="string" name="collectionMode" managedBy="user" function="validValues(['REFRESH','INCREMENTAL'])"></PARAM>
|
28
|
<PARAM required="true" type="string" name="metadataEncoding" managedBy="user">XML</PARAM>
|
29
|
<PARAM managedBy="user" name="fromDateOverride" required="false" type="string"/>
|
30
|
<PARAM managedBy="user" name="maxNumberOfRetry" required="false" type="int">5</PARAM>
|
31
|
<PARAM managedBy="user" name="requestDelay" required="false" type="int">0</PARAM>
|
32
|
<PARAM managedBy="user" name="retryDelay" required="false" type="int">60</PARAM>
|
33
|
<PARAM managedBy="user" name="connectTimeOut" required="false" type="int">30</PARAM>
|
34
|
<PARAM managedBy="user" name="readTimeOut" required="false" type="int">60</PARAM>
|
35
|
</PARAMETERS>
|
36
|
<ARCS>
|
37
|
<ARC to="COLLECT_HADOOP"/>
|
38
|
</ARCS>
|
39
|
</NODE>
|
40
|
|
41
|
<NODE name="COLLECT_HADOOP" type="SubmitDnetHadoopJobNode">
|
42
|
<DESCRIPTION>Start the Hadoop Job</DESCRIPTION>
|
43
|
<PARAMETERS>
|
44
|
<PARAM managedBy="system" name="hadoopJob" required="true" type="string">executeOozieJob</PARAM>
|
45
|
<PARAM managedBy="user" name="cluster" required="true" type="string">DHP</PARAM>
|
46
|
<PARAM managedBy="system" name="envParams" required="true" type="string">
|
47
|
{
|
48
|
"apiDescription":"apiDescription",
|
49
|
"dataSourceInfo":"dataSourceInfo",
|
50
|
"identifierPath":"identifierPath",
|
51
|
"metadataEncoding":"metadataEncoding",
|
52
|
"timestamp":"timestamp",
|
53
|
"workflowId":"workflowId",
|
54
|
"mdStoreID":"mdId",
|
55
|
"collectionMode":"collectionMode",
|
56
|
"maxNumberOfRetry":"maxNumberOfRetry",
|
57
|
"requestDelay":"requestDelay",
|
58
|
"retryDelay":"retryDelay",
|
59
|
"connectTimeOut":"connectTimeOut",
|
60
|
"readTimeOut":"readTimeOut",
|
61
|
"dnetMessageManagerURL":"dnetMessageManagerURL",
|
62
|
"oozie.wf.application.path":"oozieWfPath"
|
63
|
}
|
64
|
</PARAM>
|
65
|
<PARAM managedBy="system" name="params" required="true" type="string">
|
66
|
{
|
67
|
"collection_java_xmx" : "-Xmx300m"
|
68
|
}
|
69
|
</PARAM
|
70
|
<PARAM managedBy="system" name="oozieReportActionsCsv" required="true" type="string">BeginRead,StartTransaction,CollectionWorker</PARAM>
|
71
|
</PARAMETERS>
|
72
|
<ARCS>
|
73
|
<ARC to="UPDATE_INFO"/>
|
74
|
</ARCS>
|
75
|
</NODE>
|
76
|
|
77
|
<NODE name="UPDATE_INFO" type="MDStoreToApiExtraFieldHadoop">
|
78
|
<DESCRIPTION>Update datasouce API extra fields</DESCRIPTION>
|
79
|
<PARAMETERS>
|
80
|
<PARAM required="true" type="string" name="mdId" managedBy="system">$params.("harv_id")$</PARAM>
|
81
|
<PARAM required="true" type="string" name="datasourceId" managedBy="system">$params.("dataprovider:id")$</PARAM>
|
82
|
<PARAM required="true" type="string" name="datasourceInterface" managedBy="system">$params.("dataprovider:interface")$</PARAM>
|
83
|
<PARAM required="true" type="string" name="extraFieldForTotal" managedBy="system">last_collection_total</PARAM>
|
84
|
<PARAM required="true" type="string" name="extraFieldForDate" managedBy="system">last_collection_date</PARAM>
|
85
|
<PARAM required="true" type="string" name="extraFieldForMdId" managedBy="system">last_collection_mdId</PARAM>
|
86
|
</PARAMETERS>
|
87
|
<ARCS>
|
88
|
<ARC to="success"/>
|
89
|
</ARCS>
|
90
|
</NODE>
|