Revision 28894
Added by Claudio Atzori almost 10 years ago
| modules/dnet-openaireplus-workflows/trunk/src/main/resources/eu/dnetlib/test/profiles/openaireplus/workflows/provision/index.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="35044412-9bee-479b-a8d5-e15ba2fcad3a_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>Index from HBASE</WORKFLOW_NAME> |
|
| 12 |
<WORKFLOW_TYPE>Provision</WORKFLOW_TYPE> |
|
| 13 |
<WORKFLOW_PRIORITY>30</WORKFLOW_PRIORITY> |
|
| 14 |
<CONFIGURATION start="manual"> |
|
| 15 |
<NODE name="fetchRelClasses" type="FetchRelClasses" isStart="true"> |
|
| 16 |
<DESCRIPTION /> |
|
| 17 |
<PARAMETERS> |
|
| 18 |
<PARAM name="relClassesProperty" type="string" managedBy="system" required="true">dnet.openaire.model.relclasses.xquery</PARAM> |
|
| 19 |
<PARAM name="relClassesName" type="string" managedBy="system" required="true">relClasses</PARAM> |
|
| 20 |
</PARAMETERS> |
|
| 21 |
<ARCS> |
|
| 22 |
<ARC to="group" /> |
|
| 23 |
</ARCS> |
|
| 24 |
</NODE> |
|
| 25 |
<NODE name="findIndex" type="FindIndex" isStart="true"> |
|
| 26 |
<DESCRIPTION /> |
|
| 27 |
<PARAMETERS> |
|
| 28 |
<PARAM name="mdFormat" type="string" managedBy="user" required="true">DMF</PARAM> |
|
| 29 |
<PARAM name="layout" type="string" managedBy="system" required="true">index</PARAM> |
|
| 30 |
<PARAM name="interpretation" type="string" managedBy="system" required="true">openaire</PARAM> |
|
| 31 |
</PARAMETERS> |
|
| 32 |
<ARCS> |
|
| 33 |
<ARC name="found" to="prepare" /> |
|
| 34 |
<ARC name="notFound" to="createIndex" /> |
|
| 35 |
</ARCS> |
|
| 36 |
</NODE> |
|
| 37 |
<NODE name="createIndex" type="CreateIndex"> |
|
| 38 |
<DESCRIPTION /> |
|
| 39 |
<PARAMETERS /> |
|
| 40 |
<ARCS> |
|
| 41 |
<ARC to="prepare" /> |
|
| 42 |
</ARCS> |
|
| 43 |
</NODE> |
|
| 44 |
<NODE name="prepare" type="PrepareIndexJob"> |
|
| 45 |
<DESCRIPTION>Prepare indexing</DESCRIPTION> |
|
| 46 |
<PARAMETERS> |
|
| 47 |
<PARAM name="outputRecordsPathParam" type="string" managedBy="system" required="true">outputRecordsPath</PARAM> |
|
| 48 |
<PARAM name="rottenRecordsPathParam" type="string" managedBy="system" required="true">rottenRecordsPath</PARAM> |
|
| 49 |
<PARAM name="layoutToRecordStylesheet" type="string" managedBy="system" required="true">/eu/dnetlib/msro/openaireplus/workflows/index/openaireLayoutToRecordStylesheet.xsl</PARAM> |
|
| 50 |
<PARAM name="oafSchemaLocationProperty" type="string" managedBy="system" required="true">oaf.schema.location</PARAM> |
|
| 51 |
</PARAMETERS> |
|
| 52 |
<ARCS> |
|
| 53 |
<ARC to="cleanupXml" /> |
|
| 54 |
<ARC to="cleanupRotten" /> |
|
| 55 |
</ARCS> |
|
| 56 |
</NODE> |
|
| 57 |
<NODE name="cleanupXml" type="DeleteHdfsPathJob"> |
|
| 58 |
<DESCRIPTION>hdfs cleanup (xml)</DESCRIPTION> |
|
| 59 |
<PARAMETERS> |
|
| 60 |
<PARAM required="true" type="string" name="cluster" managedBy="system">DM</PARAM> |
|
| 61 |
<PARAM required="true" type="string" name="envParams" managedBy="system"> |
|
| 62 |
{
|
|
| 63 |
'path' : 'outputRecordsPath' |
|
| 64 |
} |
|
| 65 |
</PARAM> |
|
| 66 |
</PARAMETERS> |
|
| 67 |
<ARCS> |
|
| 68 |
<ARC to="group" /> |
|
| 69 |
</ARCS> |
|
| 70 |
</NODE> |
|
| 71 |
<NODE name="cleanupRotten" type="DeleteHdfsPathJob"> |
|
| 72 |
<DESCRIPTION>hdfs cleanup (rotten)</DESCRIPTION> |
|
| 73 |
<PARAMETERS> |
|
| 74 |
<PARAM required="true" type="string" name="cluster" managedBy="system">DM</PARAM> |
|
| 75 |
<PARAM required="true" type="string" name="envParams" managedBy="system"> |
|
| 76 |
{
|
|
| 77 |
'path' : 'rottenRecordsPath' |
|
| 78 |
} |
|
| 79 |
</PARAM> |
|
| 80 |
</PARAMETERS> |
|
| 81 |
<ARCS> |
|
| 82 |
<ARC to="group" /> |
|
| 83 |
<ARC to="updateStats" /> |
|
| 84 |
</ARCS> |
|
| 85 |
</NODE> |
|
| 86 |
<NODE name="group" type="SubmitHadoopJob" isJoin="true"> |
|
| 87 |
<DESCRIPTION>M/R group entities</DESCRIPTION> |
|
| 88 |
<PARAMETERS> |
|
| 89 |
<PARAM required="true" type="string" name="cluster" managedBy="system">DM</PARAM> |
|
| 90 |
<PARAM required="true" type="string" name="hadoopJob" managedBy="system">prepareIndexDataJob</PARAM> |
|
| 91 |
<PARAM required="true" type="string" name="sysParams" managedBy="system"> |
|
| 92 |
{
|
|
| 93 |
'hbase.mapred.inputtable' : 'hbase.mapred.datatable', |
|
| 94 |
'hbase.mapreduce.inputtable' : 'hbase.mapred.datatable' |
|
| 95 |
} |
|
| 96 |
</PARAM> |
|
| 97 |
<PARAM required="true" type="string" name="envParams" managedBy="system"> |
|
| 98 |
{
|
|
| 99 |
'mapred.output.dir' : 'outputRecordsPath', |
|
| 100 |
'index.entity.links' : 'index.entity.links', |
|
| 101 |
'oaf.schema.location' : 'oaf.schema.location', |
|
| 102 |
'contextmap' : 'contextmap', |
|
| 103 |
'relClasses' : 'relClasses' |
|
| 104 |
} |
|
| 105 |
</PARAM> |
|
| 106 |
</PARAMETERS> |
|
| 107 |
<ARCS> |
|
| 108 |
<ARC to="updateIndex" /> |
|
| 109 |
</ARCS> |
|
| 110 |
</NODE> |
|
| 111 |
<NODE name="updateIndex" type="SubmitHadoopJob"> |
|
| 112 |
<DESCRIPTION>M/O index records</DESCRIPTION> |
|
| 113 |
<PARAMETERS> |
|
| 114 |
<PARAM required="true" type="string" name="cluster" managedBy="system">DM</PARAM> |
|
| 115 |
<PARAM required="true" type="string" name="hadoopJob" managedBy="system">indexFeedJob</PARAM> |
|
| 116 |
<PARAM required="true" type="string" name="envParams" managedBy="system"> |
|
| 117 |
{
|
|
| 118 |
'mapred.input.dir' : 'outputRecordsPath', |
|
| 119 |
'mapred.output.dir' : 'rottenRecordsPath', |
|
| 120 |
'index.xslt' : 'index.xslt', |
|
| 121 |
'index.solr.url.zk' : 'index.solr.url.zk', |
|
| 122 |
'index.solr.url.list' : 'index.solr.url.list', |
|
| 123 |
'index.solr.url.local' : 'index.solr.url.local', |
|
| 124 |
'index.solr.collection' : 'index.solr.collection', |
|
| 125 |
'index.buffer.flush.threshold' : 'index.buffer.flush.threshold', |
|
| 126 |
'index.shutdown.wait.time' : 'index.shutdown.wait.time', |
|
| 127 |
'index.solr.local.feeding' : 'index.solr.local.feeding', |
|
| 128 |
'index.solr.sim.mode' : 'index.solr.sim.mode', |
|
| 129 |
'index.feed.timestamp' : 'index.feed.timestamp' |
|
| 130 |
} |
|
| 131 |
</PARAM> |
|
| 132 |
</PARAMETERS> |
|
| 133 |
<ARCS> |
|
| 134 |
<ARC to="sync" /> |
|
| 135 |
</ARCS> |
|
| 136 |
</NODE> |
|
| 137 |
<NODE name="updateStats"> |
|
| 138 |
<DESCRIPTION>update stats</DESCRIPTION> |
|
| 139 |
<PARAMETERS /> |
|
| 140 |
<ARCS> |
|
| 141 |
<ARC to="sync" /> |
|
| 142 |
</ARCS> |
|
| 143 |
</NODE> |
|
| 144 |
<NODE name="sync" isJoin="true"> |
|
| 145 |
<DESCRIPTION /> |
|
| 146 |
<PARAMETERS /> |
|
| 147 |
<ARCS> |
|
| 148 |
<ARC to="finalize" /> |
|
| 149 |
<ARC to="findSearchService" /> |
|
| 150 |
</ARCS> |
|
| 151 |
</NODE> |
|
| 152 |
<NODE name="finalize" type="FinalizeIndexFeeding"> |
|
| 153 |
<DESCRIPTION>commit changes</DESCRIPTION> |
|
| 154 |
<PARAMETERS /> |
|
| 155 |
<ARCS> |
|
| 156 |
<ARC to="updateDs" /> |
|
| 157 |
</ARCS> |
|
| 158 |
</NODE> |
|
| 159 |
<NODE name="updateDs" type="IndexDsUpdateJob"> |
|
| 160 |
<DESCRIPTION>update DS</DESCRIPTION> |
|
| 161 |
<PARAMETERS /> |
|
| 162 |
<ARCS> |
|
| 163 |
<ARC to="success" /> |
|
| 164 |
</ARCS> |
|
| 165 |
</NODE> |
|
| 166 |
<NODE name="findSearchService" type="FindSearchService"> |
|
| 167 |
<DESCRIPTION>find search service</DESCRIPTION> |
|
| 168 |
<PARAMETERS /> |
|
| 169 |
<ARCS> |
|
| 170 |
<ARC name="found" to="switch" /> |
|
| 171 |
<ARC name="notFound" to="success" /> |
|
| 172 |
</ARCS> |
|
| 173 |
</NODE> |
|
| 174 |
<NODE name="switch" type="SwitchIndex"> |
|
| 175 |
<DESCRIPTION>switch index</DESCRIPTION> |
|
| 176 |
<PARAMETERS /> |
|
| 177 |
<ARCS> |
|
| 178 |
<ARC to="success" /> |
|
| 179 |
</ARCS> |
|
| 180 |
</NODE> |
|
| 181 |
</CONFIGURATION> |
|
| 182 |
<STATUS /> |
|
| 183 |
</BODY> |
|
| 184 |
</RESOURCE_PROFILE> |
|
| 185 |
|
|
| 186 |
|
|
| modules/dnet-openaireplus-workflows/trunk/src/main/resources/eu/dnetlib/test/profiles/openaireplus/workflows/provision/index.hbase.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="35044412-9bee-479b-a8d5-e15ba2fcad3a_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>Index from HBASE</WORKFLOW_NAME> |
|
| 12 |
<WORKFLOW_TYPE>Provision</WORKFLOW_TYPE> |
|
| 13 |
<WORKFLOW_PRIORITY>30</WORKFLOW_PRIORITY> |
|
| 14 |
<CONFIGURATION start="manual"> |
|
| 15 |
<NODE name="fetchRelClasses" type="FetchRelClasses" isStart="true"> |
|
| 16 |
<DESCRIPTION /> |
|
| 17 |
<PARAMETERS> |
|
| 18 |
<PARAM name="relClassesProperty" type="string" managedBy="system" required="true">dnet.openaire.model.relclasses.xquery</PARAM> |
|
| 19 |
<PARAM name="relClassesName" type="string" managedBy="system" required="true">relClasses</PARAM> |
|
| 20 |
</PARAMETERS> |
|
| 21 |
<ARCS> |
|
| 22 |
<ARC to="group" /> |
|
| 23 |
</ARCS> |
|
| 24 |
</NODE> |
|
| 25 |
<NODE name="findIndex" type="FindIndex" isStart="true"> |
|
| 26 |
<DESCRIPTION /> |
|
| 27 |
<PARAMETERS> |
|
| 28 |
<PARAM name="mdFormat" type="string" managedBy="user" required="true">DMF</PARAM> |
|
| 29 |
<PARAM name="layout" type="string" managedBy="system" required="true">index</PARAM> |
|
| 30 |
<PARAM name="interpretation" type="string" managedBy="system" required="true">openaire</PARAM> |
|
| 31 |
</PARAMETERS> |
|
| 32 |
<ARCS> |
|
| 33 |
<ARC name="found" to="prepare" /> |
|
| 34 |
<ARC name="notFound" to="createIndex" /> |
|
| 35 |
</ARCS> |
|
| 36 |
</NODE> |
|
| 37 |
<NODE name="createIndex" type="CreateIndex"> |
|
| 38 |
<DESCRIPTION /> |
|
| 39 |
<PARAMETERS /> |
|
| 40 |
<ARCS> |
|
| 41 |
<ARC to="prepare" /> |
|
| 42 |
</ARCS> |
|
| 43 |
</NODE> |
|
| 44 |
<NODE name="prepare" type="PrepareIndexJob"> |
|
| 45 |
<DESCRIPTION>Prepare indexing</DESCRIPTION> |
|
| 46 |
<PARAMETERS> |
|
| 47 |
<PARAM name="outputRecordsPathParam" type="string" managedBy="system" required="true">outputRecordsPath</PARAM> |
|
| 48 |
<PARAM name="rottenRecordsPathParam" type="string" managedBy="system" required="true">rottenRecordsPath</PARAM> |
|
| 49 |
<PARAM name="layoutToRecordStylesheet" type="string" managedBy="system" required="true">/eu/dnetlib/msro/openaireplus/workflows/index/openaireLayoutToRecordStylesheet.xsl</PARAM> |
|
| 50 |
<PARAM name="oafSchemaLocationProperty" type="string" managedBy="system" required="true">oaf.schema.location</PARAM> |
|
| 51 |
</PARAMETERS> |
|
| 52 |
<ARCS> |
|
| 53 |
<ARC to="cleanupXml" /> |
|
| 54 |
<ARC to="cleanupRotten" /> |
|
| 55 |
</ARCS> |
|
| 56 |
</NODE> |
|
| 57 |
<NODE name="cleanupXml" type="DeleteHdfsPathJob"> |
|
| 58 |
<DESCRIPTION>hdfs cleanup (xml)</DESCRIPTION> |
|
| 59 |
<PARAMETERS> |
|
| 60 |
<PARAM required="true" type="string" name="cluster" managedBy="system">DM</PARAM> |
|
| 61 |
<PARAM required="true" type="string" name="envParams" managedBy="system"> |
|
| 62 |
{
|
|
| 63 |
'path' : 'outputRecordsPath' |
|
| 64 |
} |
|
| 65 |
</PARAM> |
|
| 66 |
</PARAMETERS> |
|
| 67 |
<ARCS> |
|
| 68 |
<ARC to="group" /> |
|
| 69 |
</ARCS> |
|
| 70 |
</NODE> |
|
| 71 |
<NODE name="cleanupRotten" type="DeleteHdfsPathJob"> |
|
| 72 |
<DESCRIPTION>hdfs cleanup (rotten)</DESCRIPTION> |
|
| 73 |
<PARAMETERS> |
|
| 74 |
<PARAM required="true" type="string" name="cluster" managedBy="system">DM</PARAM> |
|
| 75 |
<PARAM required="true" type="string" name="envParams" managedBy="system"> |
|
| 76 |
{
|
|
| 77 |
'path' : 'rottenRecordsPath' |
|
| 78 |
} |
|
| 79 |
</PARAM> |
|
| 80 |
</PARAMETERS> |
|
| 81 |
<ARCS> |
|
| 82 |
<ARC to="group" /> |
|
| 83 |
<ARC to="updateStats" /> |
|
| 84 |
</ARCS> |
|
| 85 |
</NODE> |
|
| 86 |
<NODE name="group" type="SubmitHadoopJob" isJoin="true"> |
|
| 87 |
<DESCRIPTION>M/R group entities</DESCRIPTION> |
|
| 88 |
<PARAMETERS> |
|
| 89 |
<PARAM required="true" type="string" name="cluster" managedBy="system">DM</PARAM> |
|
| 90 |
<PARAM required="true" type="string" name="hadoopJob" managedBy="system">prepareIndexDataJob</PARAM> |
|
| 91 |
<PARAM required="true" type="string" name="sysParams" managedBy="system"> |
|
| 92 |
{
|
|
| 93 |
'hbase.mapred.inputtable' : 'hbase.mapred.datatable', |
|
| 94 |
'hbase.mapreduce.inputtable' : 'hbase.mapred.datatable' |
|
| 95 |
} |
|
| 96 |
</PARAM> |
|
| 97 |
<PARAM required="true" type="string" name="envParams" managedBy="system"> |
|
| 98 |
{
|
|
| 99 |
'mapred.output.dir' : 'outputRecordsPath', |
|
| 100 |
'index.entity.links' : 'index.entity.links', |
|
| 101 |
'oaf.schema.location' : 'oaf.schema.location', |
|
| 102 |
'contextmap' : 'contextmap', |
|
| 103 |
'relClasses' : 'relClasses' |
|
| 104 |
} |
|
| 105 |
</PARAM> |
|
| 106 |
</PARAMETERS> |
|
| 107 |
<ARCS> |
|
| 108 |
<ARC to="updateIndex" /> |
|
| 109 |
</ARCS> |
|
| 110 |
</NODE> |
|
| 111 |
<NODE name="updateIndex" type="SubmitHadoopJob"> |
|
| 112 |
<DESCRIPTION>M/O index records</DESCRIPTION> |
|
| 113 |
<PARAMETERS> |
|
| 114 |
<PARAM required="true" type="string" name="cluster" managedBy="system">DM</PARAM> |
|
| 115 |
<PARAM required="true" type="string" name="hadoopJob" managedBy="system">indexFeedJob</PARAM> |
|
| 116 |
<PARAM required="true" type="string" name="envParams" managedBy="system"> |
|
| 117 |
{
|
|
| 118 |
'mapred.input.dir' : 'outputRecordsPath', |
|
| 119 |
'mapred.output.dir' : 'rottenRecordsPath', |
|
| 120 |
'index.xslt' : 'index.xslt', |
|
| 121 |
'index.solr.url.zk' : 'index.solr.url.zk', |
|
| 122 |
'index.solr.url.list' : 'index.solr.url.list', |
|
| 123 |
'index.solr.url.local' : 'index.solr.url.local', |
|
| 124 |
'index.solr.collection' : 'index.solr.collection', |
|
| 125 |
'index.buffer.flush.threshold' : 'index.buffer.flush.threshold', |
|
| 126 |
'index.shutdown.wait.time' : 'index.shutdown.wait.time', |
|
| 127 |
'index.solr.local.feeding' : 'index.solr.local.feeding', |
|
| 128 |
'index.solr.sim.mode' : 'index.solr.sim.mode', |
|
| 129 |
'index.feed.timestamp' : 'index.feed.timestamp' |
|
| 130 |
} |
|
| 131 |
</PARAM> |
|
| 132 |
</PARAMETERS> |
|
| 133 |
<ARCS> |
|
| 134 |
<ARC to="sync" /> |
|
| 135 |
</ARCS> |
|
| 136 |
</NODE> |
|
| 137 |
<NODE name="updateStats"> |
|
| 138 |
<DESCRIPTION>update stats</DESCRIPTION> |
|
| 139 |
<PARAMETERS /> |
|
| 140 |
<ARCS> |
|
| 141 |
<ARC to="sync" /> |
|
| 142 |
</ARCS> |
|
| 143 |
</NODE> |
|
| 144 |
<NODE name="sync" isJoin="true"> |
|
| 145 |
<DESCRIPTION /> |
|
| 146 |
<PARAMETERS /> |
|
| 147 |
<ARCS> |
|
| 148 |
<ARC to="finalize" /> |
|
| 149 |
<ARC to="findSearchService" /> |
|
| 150 |
</ARCS> |
|
| 151 |
</NODE> |
|
| 152 |
<NODE name="finalize" type="FinalizeIndexFeeding"> |
|
| 153 |
<DESCRIPTION>commit changes</DESCRIPTION> |
|
| 154 |
<PARAMETERS /> |
|
| 155 |
<ARCS> |
|
| 156 |
<ARC to="updateDs" /> |
|
| 157 |
</ARCS> |
|
| 158 |
</NODE> |
|
| 159 |
<NODE name="updateDs" type="IndexDsUpdateJob"> |
|
| 160 |
<DESCRIPTION>update DS</DESCRIPTION> |
|
| 161 |
<PARAMETERS /> |
|
| 162 |
<ARCS> |
|
| 163 |
<ARC to="success" /> |
|
| 164 |
</ARCS> |
|
| 165 |
</NODE> |
|
| 166 |
<NODE name="findSearchService" type="FindSearchService"> |
|
| 167 |
<DESCRIPTION>find search service</DESCRIPTION> |
|
| 168 |
<PARAMETERS /> |
|
| 169 |
<ARCS> |
|
| 170 |
<ARC name="found" to="switch" /> |
|
| 171 |
<ARC name="notFound" to="success" /> |
|
| 172 |
</ARCS> |
|
| 173 |
</NODE> |
|
| 174 |
<NODE name="switch" type="SwitchIndex"> |
|
| 175 |
<DESCRIPTION>switch index</DESCRIPTION> |
|
| 176 |
<PARAMETERS /> |
|
| 177 |
<ARCS> |
|
| 178 |
<ARC to="success" /> |
|
| 179 |
</ARCS> |
|
| 180 |
</NODE> |
|
| 181 |
</CONFIGURATION> |
|
| 182 |
<STATUS /> |
|
| 183 |
</BODY> |
|
| 184 |
</RESOURCE_PROFILE> |
|
| 185 |
|
|
| 186 |
|
|
Also available in: Unified diff
renamed wf file name