diff --git a/dhp-workflows/dhp-enrichment/src/main/java/eu/dnetlib/dhp/countrypropagation/PrepareDatasourceCountryAssociation.java b/dhp-workflows/dhp-enrichment/src/main/java/eu/dnetlib/dhp/countrypropagation/PrepareDatasourceCountryAssociation.java index 2ffe6f36de..430c265924 100644 --- a/dhp-workflows/dhp-enrichment/src/main/java/eu/dnetlib/dhp/countrypropagation/PrepareDatasourceCountryAssociation.java +++ b/dhp-workflows/dhp-enrichment/src/main/java/eu/dnetlib/dhp/countrypropagation/PrepareDatasourceCountryAssociation.java @@ -90,7 +90,7 @@ public class PrepareDatasourceCountryAssociation { (FilterFunction) ds -> !ds.getDataInfo().getDeletedbyinference() && Optional.ofNullable(ds.getDatasourcetype()).isPresent() && Optional.ofNullable(ds.getDatasourcetype().getClassid()).isPresent() && - (allowedtypes.contains(ds.getDatasourcetype().getClassid()) || + (allowedtypes.contains(ds.getJurisdiction().getClassid()) || whitelist.contains(ds.getId()))); // filtering of the relations taking the non deleted by inference and those with IsProvidedBy as relclass diff --git a/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/bulktag/datasourcemaster_parameters.json b/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/bulktag/datasourcemaster_parameters.json deleted file mode 100644 index 9a2eadaa7d..0000000000 --- a/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/bulktag/datasourcemaster_parameters.json +++ /dev/null @@ -1,32 +0,0 @@ -[ - { - "paramName": "p", - "paramLongName": "hdfsPath", - "paramDescription": "the path where storing the sequential file", - "paramRequired": true - }, - { - "paramName": "nn", - "paramLongName": "hdfsNameNode", - "paramDescription": "the name node on hdfs", - "paramRequired": true - }, - { - "paramName": "pgurl", - "paramLongName": "postgresUrl", - "paramDescription": "postgres url, example: jdbc:postgresql://localhost:5432/testdb", - "paramRequired": true - }, - { - "paramName": "pguser", - "paramLongName": "postgresUser", - "paramDescription": "postgres user", - "paramRequired": false - }, - { - "paramName": "pgpasswd", - "paramLongName": "postgresPassword", - "paramDescription": "postgres password", - "paramRequired": false - } -] \ No newline at end of file diff --git a/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/bulktag/input_bulkTag_parameters.json b/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/bulktag/input_bulkTag_parameters.json deleted file mode 100644 index ce1a8ecab6..0000000000 --- a/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/bulktag/input_bulkTag_parameters.json +++ /dev/null @@ -1,38 +0,0 @@ -[ - { - "paramName":"s", - "paramLongName":"sourcePath", - "paramDescription": "the path of the sequencial file to read", - "paramRequired": true - }, - { - "paramName": "pm", - "paramLongName":"pathMap", - "paramDescription": "the json path associated to each selection field", - "paramRequired": true - }, - { - "paramName": "out", - "paramLongName": "outputPath", - "paramDescription": "the path used to store temporary output files", - "paramRequired": true - }, - { - "paramName": "ssm", - "paramLongName": "isSparkSessionManaged", - "paramDescription": "true if the spark session is managed, false otherwise", - "paramRequired": false - }, - { - "paramName": "tg", - "paramLongName": "taggingConf", - "paramDescription": "this parameter is intended for testing purposes only. It is a possible tagging configuration obtained via the XQUERY. Intended to be removed", - "paramRequired": false - }, - { - "paramName": "bu", - "paramLongName": "baseURL", - "paramDescription": "this parameter is to specify the api to be queried (beta or production)", - "paramRequired": false - } -] \ No newline at end of file diff --git a/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/bulktag/input_eoscTag_parameters.json b/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/bulktag/input_eoscTag_parameters.json deleted file mode 100644 index 4c25fea019..0000000000 --- a/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/bulktag/input_eoscTag_parameters.json +++ /dev/null @@ -1,21 +0,0 @@ -[ - { - "paramName":"s", - "paramLongName":"sourcePath", - "paramDescription": "the path of the sequencial file to read", - "paramRequired": true - }, - { - "paramName": "wp", - "paramLongName": "workingPath", - "paramDescription": "the path used to store temporary output files", - "paramRequired": true - }, - { - "paramName": "ssm", - "paramLongName": "isSparkSessionManaged", - "paramDescription": "true if the spark session is managed, false otherwise", - "paramRequired": false - } - -] \ No newline at end of file diff --git a/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/bulktag/input_eosc_bulkTag_parameters.json b/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/bulktag/input_eosc_bulkTag_parameters.json deleted file mode 100644 index 5aace346d9..0000000000 --- a/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/bulktag/input_eosc_bulkTag_parameters.json +++ /dev/null @@ -1,41 +0,0 @@ -[ - - { - "paramName":"s", - "paramLongName":"sourcePath", - "paramDescription": "the path of the sequencial file to read", - "paramRequired": true - }, - { - "paramName": "dmp", - "paramLongName":"datasourceMapPath", - "paramDescription": "the path where the association datasource master has been stored", - "paramRequired": true - }, - { - "paramName":"tn", - "paramLongName":"resultTableName", - "paramDescription": "the name of the result table we are currently working on", - "paramRequired": true - }, - { - "paramName": "wp", - "paramLongName": "workingPath", - "paramDescription": "the path used to store temporary output files", - "paramRequired": true - }, - { - "paramName": "ssm", - "paramLongName": "isSparkSessionManaged", - "paramDescription": "true if the spark session is managed, false otherwise", - "paramRequired": false - }, - { - - "paramName": "rt", - "paramLongName": "resultType", - "paramDescription": "the result type", - "paramRequired": true - } - -] \ No newline at end of file diff --git a/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/bulktag/oozie_app/config-default.xml b/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/bulktag/oozie_app/config-default.xml deleted file mode 100644 index fe82ae1940..0000000000 --- a/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/bulktag/oozie_app/config-default.xml +++ /dev/null @@ -1,54 +0,0 @@ - - - jobTracker - yarnRM - - - nameNode - hdfs://nameservice1 - - - oozie.use.system.libpath - true - - - oozie.action.sharelib.for.spark - spark2 - - - hive_metastore_uris - thrift://iis-cdh5-test-m3.ocean.icm.edu.pl:9083 - - - spark2YarnHistoryServerAddress - http://iis-cdh5-test-gw.ocean.icm.edu.pl:18089 - - - spark2ExtraListeners - com.cloudera.spark.lineage.NavigatorAppListener - - - spark2SqlQueryExecutionListeners - com.cloudera.spark.lineage.NavigatorQueryListener - - - sparkExecutorNumber - 4 - - - spark2EventLogDir - /user/spark/spark2ApplicationHistory - - - sparkDriverMemory - 15G - - - sparkExecutorMemory - 6G - - - sparkExecutorCores - 1 - - \ No newline at end of file diff --git a/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/bulktag/oozie_app/workflow.xml b/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/bulktag/oozie_app/workflow.xml deleted file mode 100644 index 0d4d1f046d..0000000000 --- a/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/bulktag/oozie_app/workflow.xml +++ /dev/null @@ -1,120 +0,0 @@ - - - - sourcePath - the source path - - - pathMap - the json path associated to each selection field - - - outputPath - the output path - - - baseURL - the community API base URL - - - - - ${jobTracker} - ${nameNode} - - - oozie.action.sharelib.for.spark - ${oozieActionShareLibForSpark2} - - - - - - - - Action failed, error message[${wf:errorMessage(wf:lastErrorNode())}] - - - - - - - - - - - - - - - - - - - - - ${nameNode}/${sourcePath}/relation - ${nameNode}/${outputPath}/relation - - - - - - - - ${nameNode}/${sourcePath}/organization - ${nameNode}/${outputPath}/organization - - - - - - - - ${nameNode}/${sourcePath}/project - ${nameNode}/${outputPath}/project - - - - - - - - ${nameNode}/${sourcePath}/datasource - ${nameNode}/${outputPath}/datasource - - - - - - - - - - yarn-cluster - cluster - bulkTagging-result - eu.dnetlib.dhp.bulktag.SparkBulkTagJob - dhp-enrichment-${projectVersion}.jar - - --num-executors=${sparkExecutorNumber} - --executor-memory=${sparkExecutorMemory} - --executor-cores=${sparkExecutorCores} - --driver-memory=${sparkDriverMemory} - --conf spark.extraListeners=${spark2ExtraListeners} - --conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners} - --conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress} - --conf spark.eventLog.dir=${nameNode}${spark2EventLogDir} - - --sourcePath${sourcePath}/ - --outputPath${outputPath}/ - --pathMap${pathMap} - --baseURL${baseURL} - - - - - - - - \ No newline at end of file diff --git a/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/countrypropagation/input_countrypropagation_parameters.json b/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/countrypropagation/input_countrypropagation_parameters.json deleted file mode 100644 index f217e24582..0000000000 --- a/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/countrypropagation/input_countrypropagation_parameters.json +++ /dev/null @@ -1,32 +0,0 @@ -[ - { - "paramName":"s", - "paramLongName":"sourcePath", - "paramDescription": "the path of the sequencial file to read", - "paramRequired": true - }, - { - "paramName":"tn", - "paramLongName":"resultTableName", - "paramDescription": "the name of the result table we are currently working on", - "paramRequired": true - }, - { - "paramName": "out", - "paramLongName": "outputPath", - "paramDescription": "the path used to store temporary output files", - "paramRequired": true - }, - { - "paramName": "p", - "paramLongName": "preparedInfoPath", - "paramDescription": "the path where prepared info have been stored", - "paramRequired": false - }, - { - "paramName": "ssm", - "paramLongName": "isSparkSessionManaged", - "paramDescription": "true if the spark session is managed, false otherwise", - "paramRequired": false - } -] \ No newline at end of file diff --git a/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/countrypropagation/input_prepareassoc_parameters.json b/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/countrypropagation/input_prepareassoc_parameters.json deleted file mode 100644 index a00105f2ba..0000000000 --- a/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/countrypropagation/input_prepareassoc_parameters.json +++ /dev/null @@ -1,32 +0,0 @@ -[ - { - "paramName":"s", - "paramLongName":"sourcePath", - "paramDescription": "the path of the sequencial file to read", - "paramRequired": true - }, - { - "paramName": "out", - "paramLongName": "outputPath", - "paramDescription": "the path used to store temporary output files", - "paramRequired": true - }, - { - "paramName": "w", - "paramLongName": "whitelist", - "paramDescription": "the datasource having a type different from the allowed ones but that we want to add anyway", - "paramRequired": true - }, - { - "paramName": "at", - "paramLongName": "allowedtypes", - "paramDescription": "the allowed datasource types for country propagation", - "paramRequired": true - }, - { - "paramName": "ssm", - "paramLongName": "isSparkSessionManaged", - "paramDescription": "true if the spark session is managed, false otherwise", - "paramRequired": false - } -] \ No newline at end of file diff --git a/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/countrypropagation/input_prepareresultcountry_parameters.json b/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/countrypropagation/input_prepareresultcountry_parameters.json deleted file mode 100644 index 18163d1f96..0000000000 --- a/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/countrypropagation/input_prepareresultcountry_parameters.json +++ /dev/null @@ -1,38 +0,0 @@ -[ - { - "paramName":"s", - "paramLongName":"sourcePath", - "paramDescription": "the path of the sequencial file to read", - "paramRequired": true - }, - { - "paramName":"out", - "paramLongName":"outputPath", - "paramDescription": "the output path", - "paramRequired": true - }, - { - "paramName":"w", - "paramLongName":"workingPath", - "paramDescription": "the working path", - "paramRequired": true - }, - { - "paramName":"tn", - "paramLongName":"resultTableName", - "paramDescription": "the name of the result table we are currently working on", - "paramRequired": true - }, - { - "paramName": "p", - "paramLongName": "preparedInfoPath", - "paramDescription": "the path where prepared info have been stored", - "paramRequired": true - }, - { - "paramName": "ssm", - "paramLongName": "isSparkSessionManaged", - "paramDescription": "true if the spark session is managed, false otherwise", - "paramRequired": false - } -] \ No newline at end of file diff --git a/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/countrypropagation/oozie_app/config-default.xml b/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/countrypropagation/oozie_app/config-default.xml deleted file mode 100644 index 2744ea92ba..0000000000 --- a/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/countrypropagation/oozie_app/config-default.xml +++ /dev/null @@ -1,58 +0,0 @@ - - - jobTracker - yarnRM - - - nameNode - hdfs://nameservice1 - - - oozie.use.system.libpath - true - - - oozie.action.sharelib.for.spark - spark2 - - - hive_metastore_uris - thrift://iis-cdh5-test-m3.ocean.icm.edu.pl:9083 - - - spark2YarnHistoryServerAddress - http://iis-cdh5-test-gw.ocean.icm.edu.pl:18089 - - - spark2EventLogDir - /user/spark/spark2ApplicationHistory - - - spark2ExtraListeners - com.cloudera.spark.lineage.NavigatorAppListener - - - spark2SqlQueryExecutionListeners - com.cloudera.spark.lineage.NavigatorQueryListener - - - sparkExecutorNumber - 4 - - - sparkDriverMemory - 15G - - - sparkExecutorMemory - 6G - - - sparkExecutorCores - 1 - - - spark2MaxExecutors - 50 - - \ No newline at end of file diff --git a/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/countrypropagation/oozie_app/workflow.xml b/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/countrypropagation/oozie_app/workflow.xml deleted file mode 100644 index 271ccbf722..0000000000 --- a/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/countrypropagation/oozie_app/workflow.xml +++ /dev/null @@ -1,375 +0,0 @@ - - - - sourcePath - the source path - - - whitelist - the white list - - - allowedtypes - the allowed types - - - outputPath - the output path - - - - - - ${jobTracker} - ${nameNode} - - - oozie.action.sharelib.for.spark - ${oozieActionShareLibForSpark2} - - - - - - - - Action failed, error message[${wf:errorMessage(wf:lastErrorNode())}] - - - - - - - - - - - - - - - - - - - - - ${nameNode}/${sourcePath}/relation - ${nameNode}/${outputPath}/relation - - - - - - - - ${nameNode}/${sourcePath}/organization - ${nameNode}/${outputPath}/organization - - - - - - - - ${nameNode}/${sourcePath}/project - ${nameNode}/${outputPath}/project - - - - - - - - ${nameNode}/${sourcePath}/datasource - ${nameNode}/${outputPath}/datasource - - - - - - - - - - yarn - cluster - PrepareDatasourceCountryAssociation - eu.dnetlib.dhp.countrypropagation.PrepareDatasourceCountryAssociation - dhp-enrichment-${projectVersion}.jar - - --executor-cores=${sparkExecutorCores} - --executor-memory=${sparkExecutorMemory} - --driver-memory=${sparkDriverMemory} - --conf spark.extraListeners=${spark2ExtraListeners} - --conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners} - --conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress} - --conf spark.eventLog.dir=${nameNode}${spark2EventLogDir} - --conf spark.sql.shuffle.partitions=3840 - - --sourcePath${sourcePath} - --whitelist${whitelist} - --allowedtypes${allowedtypes} - --outputPath${workingDir}/preparedInfo - - - - - - - - - - - - - - - yarn - cluster - prepareResultCountry-Publication - eu.dnetlib.dhp.countrypropagation.PrepareResultCountrySet - dhp-enrichment-${projectVersion}.jar - - --executor-cores=${sparkExecutorCores} - --executor-memory=${sparkExecutorMemory} - --driver-memory=${sparkDriverMemory} - --conf spark.extraListeners=${spark2ExtraListeners} - --conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners} - --conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress} - --conf spark.eventLog.dir=${nameNode}${spark2EventLogDir} - --conf spark.speculation=false - --conf spark.hadoop.mapreduce.map.speculative=false - --conf spark.hadoop.mapreduce.reduce.speculative=false - --conf spark.sql.shuffle.partitions=3840 - - --sourcePath${sourcePath}/publication - --outputPath${workingDir}/publication - --workingPath${workingDir}/workingP - --resultTableNameeu.dnetlib.dhp.schema.oaf.Publication - --preparedInfoPath${workingDir}/preparedInfo - - - - - - - - yarn - cluster - prepareResultCountry-Dataset - eu.dnetlib.dhp.countrypropagation.PrepareResultCountrySet - dhp-enrichment-${projectVersion}.jar - - --executor-cores=${sparkExecutorCores} - --executor-memory=${sparkExecutorMemory} - --driver-memory=${sparkDriverMemory} - --conf spark.extraListeners=${spark2ExtraListeners} - --conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners} - --conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress} - --conf spark.eventLog.dir=${nameNode}${spark2EventLogDir} - --conf spark.speculation=false - --conf spark.hadoop.mapreduce.map.speculative=false - --conf spark.hadoop.mapreduce.reduce.speculative=false - --conf spark.sql.shuffle.partitions=3840 - - --sourcePath${sourcePath}/dataset - --outputPath${workingDir}/dataset - --workingPath${workingDir}/workingD - --resultTableNameeu.dnetlib.dhp.schema.oaf.Dataset - --preparedInfoPath${workingDir}/preparedInfo - - - - - - - - yarn - cluster - prepareResultCountry-ORP - eu.dnetlib.dhp.countrypropagation.PrepareResultCountrySet - dhp-enrichment-${projectVersion}.jar - - --executor-cores=${sparkExecutorCores} - --executor-memory=${sparkExecutorMemory} - --driver-memory=${sparkDriverMemory} - --conf spark.extraListeners=${spark2ExtraListeners} - --conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners} - --conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress} - --conf spark.eventLog.dir=${nameNode}${spark2EventLogDir} - --conf spark.speculation=false - --conf spark.hadoop.mapreduce.map.speculative=false - --conf spark.hadoop.mapreduce.reduce.speculative=false - --conf spark.sql.shuffle.partitions=3840 - - --sourcePath${sourcePath}/otherresearchproduct - --outputPath${workingDir}/otherresearchproduct - --workingPath${workingDir}/workingO - --resultTableNameeu.dnetlib.dhp.schema.oaf.OtherResearchProduct - --preparedInfoPath${workingDir}/preparedInfo - - - - - - - - yarn - cluster - prepareResultCountry-Software - eu.dnetlib.dhp.countrypropagation.PrepareResultCountrySet - dhp-enrichment-${projectVersion}.jar - - --executor-cores=${sparkExecutorCores} - --executor-memory=${sparkExecutorMemory} - --driver-memory=${sparkDriverMemory} - --conf spark.extraListeners=${spark2ExtraListeners} - --conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners} - --conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress} - --conf spark.eventLog.dir=${nameNode}${spark2EventLogDir} - --conf spark.speculation=false - --conf spark.hadoop.mapreduce.map.speculative=false - --conf spark.hadoop.mapreduce.reduce.speculative=false - --conf spark.sql.shuffle.partitions=3840 - - --sourcePath${sourcePath}/software - --outputPath${workingDir}/software - --workingPath${workingDir}/workingS - --resultTableNameeu.dnetlib.dhp.schema.oaf.Software - --preparedInfoPath${workingDir}/preparedInfo - - - - - - - - - - - - - - - - - yarn - cluster - countryPropagationForPublications - eu.dnetlib.dhp.countrypropagation.SparkCountryPropagationJob - dhp-enrichment-${projectVersion}.jar - - --executor-cores=${sparkExecutorCores} - --executor-memory=${sparkExecutorMemory} - --driver-memory=${sparkDriverMemory} - --conf spark.extraListeners=${spark2ExtraListeners} - --conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners} - --conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress} - --conf spark.eventLog.dir=${nameNode}${spark2EventLogDir} - --conf spark.speculation=false - --conf spark.hadoop.mapreduce.map.speculative=false - --conf spark.hadoop.mapreduce.reduce.speculative=false - --conf spark.sql.shuffle.partitions=3840 - - --sourcePath${sourcePath}/publication - --preparedInfoPath${workingDir}/publication - --resultTableNameeu.dnetlib.dhp.schema.oaf.Publication - --outputPath${outputPath}/publication - - - - - - - - yarn - cluster - countryPropagationForDataset - eu.dnetlib.dhp.countrypropagation.SparkCountryPropagationJob - dhp-enrichment-${projectVersion}.jar - - --executor-cores=${sparkExecutorCores} - --executor-memory=${sparkExecutorMemory} - --driver-memory=${sparkDriverMemory} - --conf spark.extraListeners=${spark2ExtraListeners} - --conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners} - --conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress} - --conf spark.eventLog.dir=${nameNode}${spark2EventLogDir} - --conf spark.speculation=false - --conf spark.hadoop.mapreduce.map.speculative=false - --conf spark.hadoop.mapreduce.reduce.speculative=false - --conf spark.sql.shuffle.partitions=3840 - - --sourcePath${sourcePath}/dataset - --preparedInfoPath${workingDir}/dataset - --resultTableNameeu.dnetlib.dhp.schema.oaf.Dataset - --outputPath${outputPath}/dataset - - - - - - - - yarn - cluster - countryPropagationForORP - eu.dnetlib.dhp.countrypropagation.SparkCountryPropagationJob - dhp-enrichment-${projectVersion}.jar - - --executor-cores=${sparkExecutorCores} - --executor-memory=${sparkExecutorMemory} - --driver-memory=${sparkDriverMemory} - --conf spark.extraListeners=${spark2ExtraListeners} - --conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners} - --conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress} - --conf spark.eventLog.dir=${nameNode}${spark2EventLogDir} - --conf spark.speculation=false - --conf spark.hadoop.mapreduce.map.speculative=false - --conf spark.hadoop.mapreduce.reduce.speculative=false - --conf spark.sql.shuffle.partitions=3840 - - --sourcePath${sourcePath}/otherresearchproduct - --preparedInfoPath${workingDir}/otherresearchproduct - --resultTableNameeu.dnetlib.dhp.schema.oaf.OtherResearchProduct - --outputPath${outputPath}/otherresearchproduct - - - - - - - - yarn - cluster - countryPropagationForSoftware - eu.dnetlib.dhp.countrypropagation.SparkCountryPropagationJob - dhp-enrichment-${projectVersion}.jar - - --executor-cores=${sparkExecutorCores} - --executor-memory=${sparkExecutorMemory} - --driver-memory=${sparkDriverMemory} - --conf spark.extraListeners=${spark2ExtraListeners} - --conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners} - --conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress} - --conf spark.eventLog.dir=${nameNode}${spark2EventLogDir} - --conf spark.speculation=false - --conf spark.hadoop.mapreduce.map.speculative=false - --conf spark.hadoop.mapreduce.reduce.speculative=false - --conf spark.sql.shuffle.partitions=3840 - - --sourcePath${sourcePath}/software - --preparedInfoPath${workingDir}/software - --resultTableNameeu.dnetlib.dhp.schema.oaf.Software - --outputPath${outputPath}/software - - - - - - - - - - \ No newline at end of file diff --git a/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/entitytoorganizationfromsemrel/input_preparation_parameter.json b/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/entitytoorganizationfromsemrel/input_preparation_parameter.json deleted file mode 100644 index b599373318..0000000000 --- a/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/entitytoorganizationfromsemrel/input_preparation_parameter.json +++ /dev/null @@ -1,50 +0,0 @@ -[ - { - "paramName":"gp", - "paramLongName":"graphPath", - "paramDescription": "the path of the sequencial file to read", - "paramRequired": true - }, - { - "paramName":"h", - "paramLongName":"hive_metastore_uris", - "paramDescription": "the hive metastore uris", - "paramRequired": true - }, - { - "paramName":"lp", - "paramLongName":"leavesPath", - "paramDescription": "true if the new version of the graph must be saved", - "paramRequired": false - }, - { - "paramName":"cp", - "paramLongName":"childParentPath", - "paramDescription": "path where to store/find association from datasource and organization", - "paramRequired": true - }, - { - "paramName":"rp", - "paramLongName":"resultOrgPath", - "paramDescription": "path where to store/find already linked results and organizations", - "paramRequired": true - }, - { - "paramName": "ssm", - "paramLongName": "isSparkSessionManaged", - "paramDescription": "the path where prepared info have been stored", - "paramRequired": false - }, - { - "paramName": "rep", - "paramLongName": "relationPath", - "paramDescription": "the path where to store the selected subset of relations", - "paramRequired": false - }, - { - "paramName": "pop", - "paramLongName": "projectOrganizationPath", - "paramDescription": "the number of iterations to be computed", - "paramRequired": true - } -] \ No newline at end of file diff --git a/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/entitytoorganizationfromsemrel/input_propagation_parameter.json b/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/entitytoorganizationfromsemrel/input_propagation_parameter.json deleted file mode 100644 index 66a7f5b2f5..0000000000 --- a/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/entitytoorganizationfromsemrel/input_propagation_parameter.json +++ /dev/null @@ -1,62 +0,0 @@ -[ - { - "paramName":"rep", - "paramLongName":"relationPath", - "paramDescription": "the path of the sequencial file to read", - "paramRequired": true - }, - { - "paramName":"h", - "paramLongName":"hive_metastore_uris", - "paramDescription": "the hive metastore uris", - "paramRequired": true - }, - { - "paramName":"lp", - "paramLongName":"leavesPath", - "paramDescription": "true if the new version of the graph must be saved", - "paramRequired": false - }, - { - "paramName":"cp", - "paramLongName":"childParentPath", - "paramDescription": "path where to store/find association from datasource and organization", - "paramRequired": true - }, - { - "paramName":"rp", - "paramLongName":"resultOrgPath", - "paramDescription": "path where to store/find already linked results and organizations", - "paramRequired": true - }, - { - "paramName": "ssm", - "paramLongName": "isSparkSessionManaged", - "paramDescription": "the path where prepared info have been stored", - "paramRequired": false - }, - { - "paramName": "wd", - "paramLongName": "workingDir", - "paramDescription": "true if it is a test running", - "paramRequired": false - }, - { - "paramName": "out", - "paramLongName": "outputPath", - "paramDescription": "the path used to store temporary output files", - "paramRequired": true - }, - { - "paramName": "it", - "paramLongName": "iterations", - "paramDescription": "the number of iterations to be computed", - "paramRequired": false - }, - { - "paramName": "pop", - "paramLongName": "projectOrganizationPath", - "paramDescription": "the number of iterations to be computed", - "paramRequired": true - } -] \ No newline at end of file diff --git a/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/entitytoorganizationfromsemrel/oozie_app/config-default.xml b/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/entitytoorganizationfromsemrel/oozie_app/config-default.xml deleted file mode 100644 index 2744ea92ba..0000000000 --- a/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/entitytoorganizationfromsemrel/oozie_app/config-default.xml +++ /dev/null @@ -1,58 +0,0 @@ - - - jobTracker - yarnRM - - - nameNode - hdfs://nameservice1 - - - oozie.use.system.libpath - true - - - oozie.action.sharelib.for.spark - spark2 - - - hive_metastore_uris - thrift://iis-cdh5-test-m3.ocean.icm.edu.pl:9083 - - - spark2YarnHistoryServerAddress - http://iis-cdh5-test-gw.ocean.icm.edu.pl:18089 - - - spark2EventLogDir - /user/spark/spark2ApplicationHistory - - - spark2ExtraListeners - com.cloudera.spark.lineage.NavigatorAppListener - - - spark2SqlQueryExecutionListeners - com.cloudera.spark.lineage.NavigatorQueryListener - - - sparkExecutorNumber - 4 - - - sparkDriverMemory - 15G - - - sparkExecutorMemory - 6G - - - sparkExecutorCores - 1 - - - spark2MaxExecutors - 50 - - \ No newline at end of file diff --git a/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/entitytoorganizationfromsemrel/oozie_app/workflow.xml b/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/entitytoorganizationfromsemrel/oozie_app/workflow.xml deleted file mode 100644 index d7335d8408..0000000000 --- a/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/entitytoorganizationfromsemrel/oozie_app/workflow.xml +++ /dev/null @@ -1,105 +0,0 @@ - - - - sourcePath - the source path - - - iterations - the number of hops to be done up on the hierarchy - - - - - - ${jobTracker} - ${nameNode} - - - oozie.action.sharelib.for.spark - ${oozieActionShareLibForSpark2} - - - - - - - - Action failed, error message[${wf:errorMessage(wf:lastErrorNode())}] - - - - - - - - - - - - - - yarn - cluster - PrepareResultProjectOrganizationAssociation - eu.dnetlib.dhp.entitytoorganizationfromsemrel.PrepareInfo - dhp-enrichment-${projectVersion}.jar - - --executor-cores=${sparkExecutorCores} - --executor-memory=${sparkExecutorMemory} - --driver-memory=${sparkDriverMemory} - --conf spark.extraListeners=${spark2ExtraListeners} - --conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners} - --conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress} - --conf spark.eventLog.dir=${nameNode}${spark2EventLogDir} - - --graphPath${sourcePath} - --hive_metastore_uris${hive_metastore_uris} - --leavesPath${workingDir}/preparedInfo/leavesPath - --childParentPath${workingDir}/preparedInfo/childParentPath - --resultOrgPath${workingDir}/preparedInfo/resultOrgPath - --projectOrganizationPath${workingDir}/preparedInfo/projectOrganizationPath - --relationPath${workingDir}/preparedInfo/relation - - - - - - - - yarn - cluster - resultProjectToOrganizationFromSemRel - eu.dnetlib.dhp.entitytoorganizationfromsemrel.SparkEntityToOrganizationFromSemRel - dhp-enrichment-${projectVersion}.jar - - --executor-cores=${sparkExecutorCores} - --executor-memory=${sparkExecutorMemory} - --driver-memory=${sparkDriverMemory} - --conf spark.extraListeners=${spark2ExtraListeners} - --conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners} - --conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress} - --conf spark.eventLog.dir=${nameNode}${spark2EventLogDir} - --conf spark.dynamicAllocation.enabled=true - --conf spark.dynamicAllocation.maxExecutors=${spark2MaxExecutors} - --conf spark.sql.shuffle.partitions=3840 - - --relationPath${workingDir}/preparedInfo/relation - --outputPath${sourcePath}/relation - --leavesPath${workingDir}/preparedInfo/leavesPath - --childParentPath${workingDir}/preparedInfo/childParentPath - --resultOrgPath${workingDir}/preparedInfo/resultOrgPath - --projectOrganizationPath${workingDir}/preparedInfo/projectOrganizationPath - --hive_metastore_uris${hive_metastore_uris} - --workingDir${workingDir}/working - --iterations${iterations} - - - - - - - - - - \ No newline at end of file diff --git a/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/orcidtoresultfromsemrel/input_orcidtoresult_parameters.json b/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/orcidtoresultfromsemrel/input_orcidtoresult_parameters.json deleted file mode 100644 index 3cbaa23bb6..0000000000 --- a/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/orcidtoresultfromsemrel/input_orcidtoresult_parameters.json +++ /dev/null @@ -1,44 +0,0 @@ -[ - { - "paramName":"s", - "paramLongName":"sourcePath", - "paramDescription": "the path of the sequencial file to read", - "paramRequired": true - }, - { - "paramName":"sg", - "paramLongName":"saveGraph", - "paramDescription": "true if the new version of the graph must be saved", - "paramRequired": false - }, - { - "paramName": "out", - "paramLongName": "outputPath", - "paramDescription": "the path used to store temporary output files", - "paramRequired": true - }, - { - "paramName": "ssm", - "paramLongName": "isSparkSessionManaged", - "paramDescription": "true if the spark session is managed, false otherwise", - "paramRequired": false - }, - { - "paramName":"tn", - "paramLongName":"resultTableName", - "paramDescription": "the name of the result table we are currently working on", - "paramRequired": true - }, - { - "paramName":"pu", - "paramLongName":"possibleUpdatesPath", - "paramDescription": "the path the the association resultId orcid author list can be found", - "paramRequired": true - }, - { - "paramName":"test", - "paramLongName":"isTest", - "paramDescription": "true if it is executing a test", - "paramRequired": false - } -] \ No newline at end of file diff --git a/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/orcidtoresultfromsemrel/input_prepareorcidtoresult_parameters.json b/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/orcidtoresultfromsemrel/input_prepareorcidtoresult_parameters.json deleted file mode 100644 index 08648d61a1..0000000000 --- a/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/orcidtoresultfromsemrel/input_prepareorcidtoresult_parameters.json +++ /dev/null @@ -1,38 +0,0 @@ -[ - { - "paramName":"s", - "paramLongName":"sourcePath", - "paramDescription": "the path of the sequencial file to read", - "paramRequired": true - }, - { - "paramName":"as", - "paramLongName":"allowedsemrels", - "paramDescription": "the allowed sematinc relations for propagation", - "paramRequired": true - }, - { - "paramName":"h", - "paramLongName":"hive_metastore_uris", - "paramDescription": "the hive metastore uris", - "paramRequired": true - }, - { - "paramName": "out", - "paramLongName": "outputPath", - "paramDescription": "the path used to store temporary output files", - "paramRequired": true - }, - { - "paramName": "ssm", - "paramLongName": "isSparkSessionManaged", - "paramDescription": "true if the spark session is managed, false otherwise", - "paramRequired": false - }, - { - "paramName":"tn", - "paramLongName":"resultTableName", - "paramDescription": "the name of the result table we are currently working on", - "paramRequired": true - } -] \ No newline at end of file diff --git a/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/orcidtoresultfromsemrel/input_prepareorcidtoresult_parameters2.json b/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/orcidtoresultfromsemrel/input_prepareorcidtoresult_parameters2.json deleted file mode 100644 index 1a67134a6b..0000000000 --- a/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/orcidtoresultfromsemrel/input_prepareorcidtoresult_parameters2.json +++ /dev/null @@ -1,20 +0,0 @@ -[ - { - "paramName":"s", - "paramLongName":"sourcePath", - "paramDescription": "the path of the sequencial file to read", - "paramRequired": true - }, - { - "paramName": "out", - "paramLongName": "outputPath", - "paramDescription": "the path used to store temporary output files", - "paramRequired": true - }, - { - "paramName": "ssm", - "paramLongName": "isSparkSessionManaged", - "paramDescription": "true if the spark session is managed, false otherwise", - "paramRequired": false - } -] \ No newline at end of file diff --git a/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/orcidtoresultfromsemrel/oozie_app/config-default.xml b/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/orcidtoresultfromsemrel/oozie_app/config-default.xml deleted file mode 100644 index 8d2c341057..0000000000 --- a/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/orcidtoresultfromsemrel/oozie_app/config-default.xml +++ /dev/null @@ -1,58 +0,0 @@ - - - jobTracker - yarnRM - - - nameNode - hdfs://nameservice1 - - - oozie.use.system.libpath - true - - - oozie.action.sharelib.for.spark - spark2 - - - spark2YarnHistoryServerAddress - http://iis-cdh5-test-gw.ocean.icm.edu.pl:18089 - - - hive_metastore_uris - thrift://iis-cdh5-test-m3.ocean.icm.edu.pl:9083 - - - spark2EventLogDir - /user/spark/spark2ApplicationHistory - - - spark2ExtraListeners - com.cloudera.spark.lineage.NavigatorAppListener - - - spark2SqlQueryExecutionListeners - com.cloudera.spark.lineage.NavigatorQueryListener - - - sparkExecutorNumber - 4 - - - sparkDriverMemory - 15G - - - sparkExecutorMemory - 6G - - - sparkExecutorCores - 1 - - - spark2MaxExecutors - 50 - - \ No newline at end of file diff --git a/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/orcidtoresultfromsemrel/oozie_app/workflow.xml b/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/orcidtoresultfromsemrel/oozie_app/workflow.xml deleted file mode 100644 index 5f52c16585..0000000000 --- a/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/orcidtoresultfromsemrel/oozie_app/workflow.xml +++ /dev/null @@ -1,371 +0,0 @@ - - - - sourcePath - the source path - - - allowedsemrels - the semantic relationships allowed for propagation - - - outputPath - the output path - - - - - - - Action failed, error message[${wf:errorMessage(wf:lastErrorNode())}] - - - - - - - - - - - - - - - - - - - - - ${jobTracker} - ${nameNode} - ${nameNode}/${sourcePath}/relation - ${nameNode}/${outputPath}/relation - - - - - - - - ${jobTracker} - ${nameNode} - ${nameNode}/${sourcePath}/organization - ${nameNode}/${outputPath}/organization - - - - - - - - ${jobTracker} - ${nameNode} - ${nameNode}/${sourcePath}/project - ${nameNode}/${outputPath}/project - - - - - - - - ${jobTracker} - ${nameNode} - ${nameNode}/${sourcePath}/datasource - ${nameNode}/${outputPath}/datasource - - - - - - - - - - - - - - - - - yarn - cluster - ORCIDPropagation-PreparePhase1-Publications - eu.dnetlib.dhp.orcidtoresultfromsemrel.PrepareResultOrcidAssociationStep1 - dhp-enrichment-${projectVersion}.jar - - --executor-cores=${sparkExecutorCores} - --executor-memory=${sparkExecutorMemory} - --driver-memory=${sparkDriverMemory} - --conf spark.extraListeners=${spark2ExtraListeners} - --conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners} - --conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress} - --conf spark.eventLog.dir=${nameNode}${spark2EventLogDir} - --conf spark.dynamicAllocation.enabled=true - --conf spark.dynamicAllocation.maxExecutors=${spark2MaxExecutors} - --conf spark.sql.shuffle.partitions=3840 - --conf spark.speculation=false - --conf spark.hadoop.mapreduce.map.speculative=false - --conf spark.hadoop.mapreduce.reduce.speculative=false - - --sourcePath${sourcePath} - --hive_metastore_uris${hive_metastore_uris} - --resultTableNameeu.dnetlib.dhp.schema.oaf.Publication - --outputPath${workingDir}/preparedInfo/targetOrcidAssoc - --allowedsemrels${allowedsemrels} - - - - - - - - yarn - cluster - ORCIDPropagation-PreparePhase1-Dataset - eu.dnetlib.dhp.orcidtoresultfromsemrel.PrepareResultOrcidAssociationStep1 - dhp-enrichment-${projectVersion}.jar - - --executor-cores=${sparkExecutorCores} - --executor-memory=${sparkExecutorMemory} - --driver-memory=${sparkDriverMemory} - --conf spark.extraListeners=${spark2ExtraListeners} - --conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners} - --conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress} - --conf spark.eventLog.dir=${nameNode}${spark2EventLogDir} - --conf spark.dynamicAllocation.enabled=true - --conf spark.dynamicAllocation.maxExecutors=${spark2MaxExecutors} - - --sourcePath${sourcePath} - --hive_metastore_uris${hive_metastore_uris} - --resultTableNameeu.dnetlib.dhp.schema.oaf.Dataset - --outputPath${workingDir}/preparedInfo/targetOrcidAssoc - --allowedsemrels${allowedsemrels} - - - - - - - - yarn - cluster - ORCIDPropagation-PreparePhase1-ORP - eu.dnetlib.dhp.orcidtoresultfromsemrel.PrepareResultOrcidAssociationStep1 - dhp-enrichment-${projectVersion}.jar - - --executor-cores=${sparkExecutorCores} - --executor-memory=${sparkExecutorMemory} - --driver-memory=${sparkDriverMemory} - --conf spark.extraListeners=${spark2ExtraListeners} - --conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners} - --conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress} - --conf spark.eventLog.dir=${nameNode}${spark2EventLogDir} - --conf spark.dynamicAllocation.enabled=true - --conf spark.dynamicAllocation.maxExecutors=${spark2MaxExecutors} - - --sourcePath${sourcePath} - --hive_metastore_uris${hive_metastore_uris} - --resultTableNameeu.dnetlib.dhp.schema.oaf.OtherResearchProduct - --outputPath${workingDir}/preparedInfo/targetOrcidAssoc - --allowedsemrels${allowedsemrels} - - - - - - - - yarn - cluster - ORCIDPropagation-PreparePhase1-Software - eu.dnetlib.dhp.orcidtoresultfromsemrel.PrepareResultOrcidAssociationStep1 - dhp-enrichment-${projectVersion}.jar - - --executor-cores=${sparkExecutorCores} - --executor-memory=${sparkExecutorMemory} - --driver-memory=${sparkDriverMemory} - --conf spark.extraListeners=${spark2ExtraListeners} - --conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners} - --conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress} - --conf spark.eventLog.dir=${nameNode}${spark2EventLogDir} - --conf spark.dynamicAllocation.enabled=true - --conf spark.dynamicAllocation.maxExecutors=${spark2MaxExecutors} - - --sourcePath${sourcePath} - --hive_metastore_uris${hive_metastore_uris} - --resultTableNameeu.dnetlib.dhp.schema.oaf.Software - --outputPath${workingDir}/preparedInfo/targetOrcidAssoc - --allowedsemrels${allowedsemrels} - - - - - - - - - - yarn - cluster - ORCIDPropagation-PreparePhase2 - eu.dnetlib.dhp.orcidtoresultfromsemrel.PrepareResultOrcidAssociationStep2 - dhp-enrichment-${projectVersion}.jar - - --executor-cores=${sparkExecutorCores} - --executor-memory=${sparkExecutorMemory} - --driver-memory=${sparkDriverMemory} - --conf spark.extraListeners=${spark2ExtraListeners} - --conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners} - --conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress} - --conf spark.eventLog.dir=${nameNode}${spark2EventLogDir} - --conf spark.dynamicAllocation.enabled=true - --conf spark.dynamicAllocation.maxExecutors=${spark2MaxExecutors} - - --sourcePath${workingDir}/preparedInfo/targetOrcidAssoc - --outputPath${workingDir}/preparedInfo/mergedOrcidAssoc - - - - - - - - - - - - - - - yarn - cluster - ORCIDPropagation-Publication - eu.dnetlib.dhp.orcidtoresultfromsemrel.SparkOrcidToResultFromSemRelJob - dhp-enrichment-${projectVersion}.jar - - --executor-cores=${sparkExecutorCores} - --executor-memory=${sparkExecutorMemory} - --driver-memory=${sparkDriverMemory} - --conf spark.extraListeners=${spark2ExtraListeners} - --conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners} - --conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress} - --conf spark.eventLog.dir=${nameNode}${spark2EventLogDir} - --conf spark.dynamicAllocation.enabled=true - --conf spark.dynamicAllocation.maxExecutors=${spark2MaxExecutors} - --conf spark.speculation=false - --conf spark.hadoop.mapreduce.map.speculative=false - --conf spark.hadoop.mapreduce.reduce.speculative=false - --conf spark.sql.shuffle.partitions=3840 - - --possibleUpdatesPath${workingDir}/preparedInfo/mergedOrcidAssoc - --sourcePath${sourcePath}/publication - --hive_metastore_uris${hive_metastore_uris} - --resultTableNameeu.dnetlib.dhp.schema.oaf.Publication - --outputPath${outputPath}/publication - - - - - - - - yarn - cluster - ORCIDPropagation-Dataset - eu.dnetlib.dhp.orcidtoresultfromsemrel.SparkOrcidToResultFromSemRelJob - dhp-enrichment-${projectVersion}.jar - - --executor-cores=${sparkExecutorCores} - --executor-memory=${sparkExecutorMemory} - --driver-memory=${sparkDriverMemory} - --conf spark.extraListeners=${spark2ExtraListeners} - --conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners} - --conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress} - --conf spark.eventLog.dir=${nameNode}${spark2EventLogDir} - --conf spark.dynamicAllocation.enabled=true - --conf spark.dynamicAllocation.maxExecutors=${spark2MaxExecutors} - --conf spark.speculation=false - --conf spark.hadoop.mapreduce.map.speculative=false - --conf spark.hadoop.mapreduce.reduce.speculative=false - - --possibleUpdatesPath${workingDir}/preparedInfo/mergedOrcidAssoc - --sourcePath${sourcePath}/dataset - --hive_metastore_uris${hive_metastore_uris} - --resultTableNameeu.dnetlib.dhp.schema.oaf.Dataset - --outputPath${outputPath}/dataset - - - - - - - - yarn - cluster - ORCIDPropagation-ORP - eu.dnetlib.dhp.orcidtoresultfromsemrel.SparkOrcidToResultFromSemRelJob - dhp-enrichment-${projectVersion}.jar - - --executor-cores=${sparkExecutorCores} - --executor-memory=${sparkExecutorMemory} - --driver-memory=${sparkDriverMemory} - --conf spark.extraListeners=${spark2ExtraListeners} - --conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners} - --conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress} - --conf spark.eventLog.dir=${nameNode}${spark2EventLogDir} - --conf spark.dynamicAllocation.enabled=true - --conf spark.dynamicAllocation.maxExecutors=${spark2MaxExecutors} - --conf spark.speculation=false - --conf spark.hadoop.mapreduce.map.speculative=false - --conf spark.hadoop.mapreduce.reduce.speculative=false - - --possibleUpdatesPath${workingDir}/preparedInfo/mergedOrcidAssoc - --sourcePath${sourcePath}/otherresearchproduct - --hive_metastore_uris${hive_metastore_uris} - --resultTableNameeu.dnetlib.dhp.schema.oaf.OtherResearchProduct - --outputPath${outputPath}/otherresearchproduct - - - - - - - - yarn - cluster - ORCIDPropagation-Software - eu.dnetlib.dhp.orcidtoresultfromsemrel.SparkOrcidToResultFromSemRelJob - dhp-enrichment-${projectVersion}.jar - - --executor-cores=${sparkExecutorCores} - --executor-memory=${sparkExecutorMemory} - --driver-memory=${sparkDriverMemory} - --conf spark.extraListeners=${spark2ExtraListeners} - --conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners} - --conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress} - --conf spark.eventLog.dir=${nameNode}${spark2EventLogDir} - --conf spark.dynamicAllocation.enabled=true - --conf spark.dynamicAllocation.maxExecutors=${spark2MaxExecutors} - --conf spark.speculation=false - --conf spark.hadoop.mapreduce.map.speculative=false - --conf spark.hadoop.mapreduce.reduce.speculative=false - - --possibleUpdatesPath${workingDir}/preparedInfo/mergedOrcidAssoc - --sourcePath${sourcePath}/software - --hive_metastore_uris${hive_metastore_uris} - --resultTableNameeu.dnetlib.dhp.schema.oaf.Software - --outputPath${outputPath}/software - - - - - - - - - - \ No newline at end of file diff --git a/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/projecttoresult/input_prepareprojecttoresult_parameters.json b/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/projecttoresult/input_prepareprojecttoresult_parameters.json deleted file mode 100644 index a70dbd6a08..0000000000 --- a/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/projecttoresult/input_prepareprojecttoresult_parameters.json +++ /dev/null @@ -1,33 +0,0 @@ -[ - { - "paramName":"s", - "paramLongName":"sourcePath", - "paramDescription": "the path of the sequencial file to read", - "paramRequired": true - }, - - { - "paramName":"asr", - "paramLongName":"allowedsemrels", - "paramDescription": "the types of the allowed datasources. Split by ;", - "paramRequired": true - }, - { - "paramName":"h", - "paramLongName":"hive_metastore_uris", - "paramDescription": "the hive metastore uris", - "paramRequired": true - }, - { - "paramName":"pu", - "paramLongName":"potentialUpdatePath", - "paramDescription": "the path of the potential updates ", - "paramRequired": true - }, - { - "paramName":"al", - "paramLongName":"alreadyLinkedPath", - "paramDescription": "the path of the already linked project result_set", - "paramRequired": true - } -] \ No newline at end of file diff --git a/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/projecttoresult/input_projecttoresult_parameters.json b/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/projecttoresult/input_projecttoresult_parameters.json deleted file mode 100644 index 7f44ba03cd..0000000000 --- a/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/projecttoresult/input_projecttoresult_parameters.json +++ /dev/null @@ -1,44 +0,0 @@ -[ - { - "paramName":"h", - "paramLongName":"hive_metastore_uris", - "paramDescription": "the hive metastore uris", - "paramRequired": true - }, - { - "paramName":"sg", - "paramLongName":"saveGraph", - "paramDescription": "true if the new version of the graph must be saved", - "paramRequired": false - }, - { - "paramName":"pu", - "paramLongName":"potentialUpdatePath", - "paramDescription": "the path of the potential updates ", - "paramRequired": true - }, - { - "paramName":"al", - "paramLongName":"alreadyLinkedPath", - "paramDescription": "the path of the already linked project result_set", - "paramRequired": true - }, - { - "paramName": "ssm", - "paramLongName": "isSparkSessionManaged", - "paramDescription": "the path where prepared info have been stored", - "paramRequired": false - }, - { - "paramName": "out", - "paramLongName": "outputPath", - "paramDescription": "the path used to store temporary output files", - "paramRequired": true - }, - { - "paramName": "test", - "paramLongName": "isTest", - "paramDescription": "true if it is a test running", - "paramRequired": false - } -] \ No newline at end of file diff --git a/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/projecttoresult/oozie_app/config-default.xml b/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/projecttoresult/oozie_app/config-default.xml deleted file mode 100644 index caf3c60500..0000000000 --- a/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/projecttoresult/oozie_app/config-default.xml +++ /dev/null @@ -1,63 +0,0 @@ - - - jobTracker - yarnRM - - - - nameNode - - hdfs://nameservice1 - - - oozie.use.system.libpath - true - - - oozie.action.sharelib.for.spark - spark2 - - - hive_metastore_uris - thrift://iis-cdh5-test-m3.ocean.icm.edu.pl:9083 - - - - spark2YarnHistoryServerAddress - http://iis-cdh5-test-gw.ocean.icm.edu.pl:18089 - - - spark2EventLogDir - /user/spark/spark2ApplicationHistory - - - spark2ExtraListeners - com.cloudera.spark.lineage.NavigatorAppListener - - - spark2SqlQueryExecutionListeners - com.cloudera.spark.lineage.NavigatorQueryListener - - - sparkExecutorNumber - 4 - - - sparkDriverMemory - 15G - - - sparkExecutorMemory - 6G - - - sparkExecutorCores - 1 - - - spark2MaxExecutors - 50 - - - - \ No newline at end of file diff --git a/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/projecttoresult/oozie_app/workflow.xml b/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/projecttoresult/oozie_app/workflow.xml deleted file mode 100644 index 9e91c06fb3..0000000000 --- a/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/projecttoresult/oozie_app/workflow.xml +++ /dev/null @@ -1,184 +0,0 @@ - - - - sourcePath - the source path - - - allowedsemrels - the allowed semantics - - - outputPath - the output path - - - - - ${jobTracker} - ${nameNode} - - - oozie.action.sharelib.for.spark - ${oozieActionShareLibForSpark2} - - - - - - - - Action failed, error message[${wf:errorMessage(wf:lastErrorNode())}] - - - - - - - - - - - - - - - - - - - - - - - - - ${nameNode}/${sourcePath}/relation - ${nameNode}/${outputPath}/relation - - - - - - - - ${nameNode}/${sourcePath}/publication - ${nameNode}/${outputPath}/publication - - - - - - - - ${nameNode}/${sourcePath}/dataset - ${nameNode}/${outputPath}/dataset - - - - - - - - ${nameNode}/${sourcePath}/otherresearchproduct - ${nameNode}/${outputPath}/otherresearchproduct - - - - - - - - ${nameNode}/${sourcePath}/software - ${nameNode}/${outputPath}/software - - - - - - - - ${nameNode}/${sourcePath}/organization - ${nameNode}/${outputPath}/organization - - - - - - - - ${nameNode}/${sourcePath}/project - ${nameNode}/${outputPath}/project - - - - - - - - ${nameNode}/${sourcePath}/datasource - ${nameNode}/${outputPath}/datasource - - - - - - - - - - yarn - cluster - PrepareProjectResultsAssociation - eu.dnetlib.dhp.projecttoresult.PrepareProjectResultsAssociation - dhp-enrichment-${projectVersion}.jar - - --executor-cores=${sparkExecutorCores} - --executor-memory=${sparkExecutorMemory} - --driver-memory=${sparkDriverMemory} - --conf spark.extraListeners=${spark2ExtraListeners} - --conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners} - --conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress} - --conf spark.eventLog.dir=${nameNode}${spark2EventLogDir} - - --sourcePath${sourcePath}/relation - --allowedsemrels${allowedsemrels} - --hive_metastore_uris${hive_metastore_uris} - --potentialUpdatePath${workingDir}/preparedInfo/potentialUpdates - --alreadyLinkedPath${workingDir}/preparedInfo/alreadyLinked - - - - - - - - yarn - cluster - ProjectToResultPropagation - eu.dnetlib.dhp.projecttoresult.SparkResultToProjectThroughSemRelJob - dhp-enrichment-${projectVersion}.jar - - --executor-cores=${sparkExecutorCores} - --executor-memory=${sparkExecutorMemory} - --driver-memory=${sparkDriverMemory} - --conf spark.extraListeners=${spark2ExtraListeners} - --conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners} - --conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress} - --conf spark.eventLog.dir=${nameNode}${spark2EventLogDir} - --conf spark.dynamicAllocation.enabled=true - --conf spark.dynamicAllocation.maxExecutors=${spark2MaxExecutors} - - --saveGraph${saveGraph} - --hive_metastore_uris${hive_metastore_uris} - --outputPath${outputPath}/relation - --potentialUpdatePath${workingDir}/preparedInfo/potentialUpdates - --alreadyLinkedPath${workingDir}/preparedInfo/alreadyLinked - - - - - - - - \ No newline at end of file diff --git a/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/resulttocommunityfromorganization/input_communitytoresult_parameters.json b/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/resulttocommunityfromorganization/input_communitytoresult_parameters.json deleted file mode 100644 index 0db8085d19..0000000000 --- a/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/resulttocommunityfromorganization/input_communitytoresult_parameters.json +++ /dev/null @@ -1,28 +0,0 @@ -[ - { - "paramName":"s", - "paramLongName":"sourcePath", - "paramDescription": "the path of the sequencial file to read", - "paramRequired": true - }, - - { - "paramName": "out", - "paramLongName": "outputPath", - "paramDescription": "the path used to store temporary output files", - "paramRequired": true - }, - { - "paramName": "ssm", - "paramLongName": "isSparkSessionManaged", - "paramDescription": "true if the spark session is managed, false otherwise", - "paramRequired": false - }, - { - "paramName": "p", - "paramLongName": "preparedInfoPath", - "paramDescription": "the path where prepared info have been stored", - "paramRequired": true - } - -] \ No newline at end of file diff --git a/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/resulttocommunityfromorganization/input_preparecommunitytoresult_parameters.json b/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/resulttocommunityfromorganization/input_preparecommunitytoresult_parameters.json deleted file mode 100644 index 3601db7acc..0000000000 --- a/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/resulttocommunityfromorganization/input_preparecommunitytoresult_parameters.json +++ /dev/null @@ -1,33 +0,0 @@ -[ - { - "paramName":"s", - "paramLongName":"sourcePath", - "paramDescription": "the path of the sequencial file to read", - "paramRequired": true - }, - { - "paramName":"h", - "paramLongName":"hive_metastore_uris", - "paramDescription": "the hive metastore uris", - "paramRequired": true - }, - { - "paramName": "ssm", - "paramLongName": "isSparkSessionManaged", - "paramDescription": "true if the spark session is managed, false otherwise", - "paramRequired": false - }, - { - "paramName": "out", - "paramLongName": "outputPath", - "paramDescription": "the path used to store temporary output files", - "paramRequired": true - }, - { - "paramName": "bu", - "paramLongName": "baseURL", - "paramDescription": "the base URL to the community API to use", - "paramRequired": false - } - -] \ No newline at end of file diff --git a/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/resulttocommunityfromorganization/oozie_app/config-default.xml b/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/resulttocommunityfromorganization/oozie_app/config-default.xml deleted file mode 100644 index 2744ea92ba..0000000000 --- a/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/resulttocommunityfromorganization/oozie_app/config-default.xml +++ /dev/null @@ -1,58 +0,0 @@ - - - jobTracker - yarnRM - - - nameNode - hdfs://nameservice1 - - - oozie.use.system.libpath - true - - - oozie.action.sharelib.for.spark - spark2 - - - hive_metastore_uris - thrift://iis-cdh5-test-m3.ocean.icm.edu.pl:9083 - - - spark2YarnHistoryServerAddress - http://iis-cdh5-test-gw.ocean.icm.edu.pl:18089 - - - spark2EventLogDir - /user/spark/spark2ApplicationHistory - - - spark2ExtraListeners - com.cloudera.spark.lineage.NavigatorAppListener - - - spark2SqlQueryExecutionListeners - com.cloudera.spark.lineage.NavigatorQueryListener - - - sparkExecutorNumber - 4 - - - sparkDriverMemory - 15G - - - sparkExecutorMemory - 6G - - - sparkExecutorCores - 1 - - - spark2MaxExecutors - 50 - - \ No newline at end of file diff --git a/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/resulttocommunityfromorganization/oozie_app/workflow.xml b/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/resulttocommunityfromorganization/oozie_app/workflow.xml deleted file mode 100644 index dfa762ac67..0000000000 --- a/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/resulttocommunityfromorganization/oozie_app/workflow.xml +++ /dev/null @@ -1,147 +0,0 @@ - - - - sourcePath - the source path - - - outputPath - the output path - - - baseURL - the community API base URL - - - - - ${jobTracker} - ${nameNode} - - - oozie.action.sharelib.for.spark - ${oozieActionShareLibForSpark2} - - - - - - - - Action failed, error message[${wf:errorMessage(wf:lastErrorNode())}] - - - - - - - - - - - - - - - - - - - - - ${nameNode}/${sourcePath}/relation - ${nameNode}/${outputPath}/relation - - - - - - - - ${nameNode}/${sourcePath}/organization - ${nameNode}/${outputPath}/organization - - - - - - - - ${nameNode}/${sourcePath}/project - ${nameNode}/${outputPath}/project - - - - - - - - ${nameNode}/${sourcePath}/datasource - ${nameNode}/${outputPath}/datasource - - - - - - - - - - yarn - cluster - Prepare-Community-Result-Organization - eu.dnetlib.dhp.resulttocommunityfromorganization.PrepareResultCommunitySet - dhp-enrichment-${projectVersion}.jar - - --executor-cores=6 - --executor-memory=5G - --conf spark.executor.memoryOverhead=3g - --conf spark.sql.shuffle.partitions=3284 - --driver-memory=${sparkDriverMemory} - --conf spark.extraListeners=${spark2ExtraListeners} - --conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners} - --conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress} - --conf spark.eventLog.dir=${nameNode}${spark2EventLogDir} - --conf spark.dynamicAllocation.maxExecutors=${spark2MaxExecutors} - - --sourcePath${sourcePath}/relation - --outputPath${workingDir}/preparedInfo/resultCommunityList - --hive_metastore_uris${hive_metastore_uris} - --baseURL${baseURL} - - - - - - - - yarn - cluster - community2resultfromorganization - eu.dnetlib.dhp.resulttocommunityfromorganization.SparkResultToCommunityFromOrganizationJob - dhp-enrichment-${projectVersion}.jar - - --executor-cores=6 - --executor-memory=5G - --conf spark.executor.memoryOverhead=3g - --conf spark.sql.shuffle.partitions=3284 - --driver-memory=${sparkDriverMemory} - --conf spark.extraListeners=${spark2ExtraListeners} - --conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners} - --conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress} - --conf spark.eventLog.dir=${nameNode}${spark2EventLogDir} - --conf spark.dynamicAllocation.maxExecutors=${spark2MaxExecutors} - - --preparedInfoPath${workingDir}/preparedInfo/resultCommunityList - --sourcePath${sourcePath}/ - --outputPath${outputPath}/ - - - - - - - - - - \ No newline at end of file diff --git a/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/resulttocommunityfromproject/input_communitytoresult_parameters.json b/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/resulttocommunityfromproject/input_communitytoresult_parameters.json deleted file mode 100644 index 0db8085d19..0000000000 --- a/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/resulttocommunityfromproject/input_communitytoresult_parameters.json +++ /dev/null @@ -1,28 +0,0 @@ -[ - { - "paramName":"s", - "paramLongName":"sourcePath", - "paramDescription": "the path of the sequencial file to read", - "paramRequired": true - }, - - { - "paramName": "out", - "paramLongName": "outputPath", - "paramDescription": "the path used to store temporary output files", - "paramRequired": true - }, - { - "paramName": "ssm", - "paramLongName": "isSparkSessionManaged", - "paramDescription": "true if the spark session is managed, false otherwise", - "paramRequired": false - }, - { - "paramName": "p", - "paramLongName": "preparedInfoPath", - "paramDescription": "the path where prepared info have been stored", - "paramRequired": true - } - -] \ No newline at end of file diff --git a/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/resulttocommunityfromproject/input_preparecommunitytoresult_parameters.json b/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/resulttocommunityfromproject/input_preparecommunitytoresult_parameters.json deleted file mode 100644 index cbc01c2d5a..0000000000 --- a/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/resulttocommunityfromproject/input_preparecommunitytoresult_parameters.json +++ /dev/null @@ -1,28 +0,0 @@ -[ - { - "paramName":"s", - "paramLongName":"sourcePath", - "paramDescription": "the path of the sequencial file to read", - "paramRequired": true - }, - - { - "paramName": "ssm", - "paramLongName": "isSparkSessionManaged", - "paramDescription": "true if the spark session is managed, false otherwise", - "paramRequired": false - }, - { - "paramName": "out", - "paramLongName": "outputPath", - "paramDescription": "the path used to store temporary output files", - "paramRequired": true - }, - { - "paramName": "bu", - "paramLongName": "baseURL", - "paramDescription": "the path used to store temporary output files", - "paramRequired": false - } - -] \ No newline at end of file diff --git a/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/resulttocommunityfromproject/oozie_app/config-default.xml b/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/resulttocommunityfromproject/oozie_app/config-default.xml deleted file mode 100644 index 2744ea92ba..0000000000 --- a/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/resulttocommunityfromproject/oozie_app/config-default.xml +++ /dev/null @@ -1,58 +0,0 @@ - - - jobTracker - yarnRM - - - nameNode - hdfs://nameservice1 - - - oozie.use.system.libpath - true - - - oozie.action.sharelib.for.spark - spark2 - - - hive_metastore_uris - thrift://iis-cdh5-test-m3.ocean.icm.edu.pl:9083 - - - spark2YarnHistoryServerAddress - http://iis-cdh5-test-gw.ocean.icm.edu.pl:18089 - - - spark2EventLogDir - /user/spark/spark2ApplicationHistory - - - spark2ExtraListeners - com.cloudera.spark.lineage.NavigatorAppListener - - - spark2SqlQueryExecutionListeners - com.cloudera.spark.lineage.NavigatorQueryListener - - - sparkExecutorNumber - 4 - - - sparkDriverMemory - 15G - - - sparkExecutorMemory - 6G - - - sparkExecutorCores - 1 - - - spark2MaxExecutors - 50 - - \ No newline at end of file diff --git a/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/resulttocommunityfromproject/oozie_app/workflow.xml b/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/resulttocommunityfromproject/oozie_app/workflow.xml deleted file mode 100644 index 21cc2d8874..0000000000 --- a/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/resulttocommunityfromproject/oozie_app/workflow.xml +++ /dev/null @@ -1,144 +0,0 @@ - - - - sourcePath - the source path - - - - outputPath - the output path - - - - - ${jobTracker} - ${nameNode} - - - oozie.action.sharelib.for.spark - ${oozieActionShareLibForSpark2} - - - - - - - - Action failed, error message[${wf:errorMessage(wf:lastErrorNode())}] - - - - - - - - - - - - - - - - - - - - - ${nameNode}/${sourcePath}/relation - ${nameNode}/${outputPath}/relation - - - - - - - - ${nameNode}/${sourcePath}/organization - ${nameNode}/${outputPath}/organization - - - - - - - - ${nameNode}/${sourcePath}/project - ${nameNode}/${outputPath}/project - - - - - - - - ${nameNode}/${sourcePath}/datasource - ${nameNode}/${outputPath}/datasource - - - - - - - - - - yarn - cluster - Prepare-Community-Result-Organization - eu.dnetlib.dhp.resulttocommunityfromproject.PrepareResultCommunitySet - dhp-enrichment-${projectVersion}.jar - - --executor-cores=6 - --executor-memory=5G - --conf spark.executor.memoryOverhead=3g - --conf spark.sql.shuffle.partitions=3284 - --driver-memory=${sparkDriverMemory} - --conf spark.extraListeners=${spark2ExtraListeners} - --conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners} - --conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress} - --conf spark.eventLog.dir=${nameNode}${spark2EventLogDir} - - --conf spark.dynamicAllocation.maxExecutors=${spark2MaxExecutors} - - --sourcePath${sourcePath}/relation - --outputPath${workingDir}/preparedInfo/resultCommunityList - --production${production} - - - - - - - - yarn - cluster - community2resultfromproject - eu.dnetlib.dhp.resulttocommunityfromproject.SparkResultToCommunityFromProject - dhp-enrichment-${projectVersion}.jar - - --executor-cores=6 - --executor-memory=5G - --conf spark.executor.memoryOverhead=3g - --conf spark.sql.shuffle.partitions=3284 - --driver-memory=${sparkDriverMemory} - --conf spark.extraListeners=${spark2ExtraListeners} - --conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners} - --conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress} - --conf spark.eventLog.dir=${nameNode}${spark2EventLogDir} - --conf spark.dynamicAllocation.maxExecutors=${spark2MaxExecutors} - - --preparedInfoPath${workingDir}/preparedInfo/resultCommunityList - --sourcePath${sourcePath}/ - --outputPath${outputPath}/ - - - - - - - - - - \ No newline at end of file diff --git a/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/resulttocommunityfromsemrel/input_communitytoresult_parameters.json b/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/resulttocommunityfromsemrel/input_communitytoresult_parameters.json deleted file mode 100644 index a40ce375ee..0000000000 --- a/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/resulttocommunityfromsemrel/input_communitytoresult_parameters.json +++ /dev/null @@ -1,52 +0,0 @@ -[ - - { - "paramName":"s", - "paramLongName":"sourcePath", - "paramDescription": "the path of the sequencial file to read", - "paramRequired": true - }, - { - "paramName":"sg", - "paramLongName":"saveGraph", - "paramDescription": "true if the new version of the graph must be saved", - "paramRequired": false - }, - { - "paramName":"h", - "paramLongName":"hive_metastore_uris", - "paramDescription": "the hive metastore uris", - "paramRequired": true - }, - { - "paramName": "ssm", - "paramLongName": "isSparkSessionManaged", - "paramDescription": "true if the spark session is managed, false otherwise", - "paramRequired": false - }, - { - "paramName": "out", - "paramLongName": "outputPath", - "paramDescription": "the path used to store temporary output files", - "paramRequired": true - }, - { - "paramName":"tn", - "paramLongName":"resultTableName", - "paramDescription": "the name of the result table we are currently working on", - "paramRequired": true - }, - { - "paramName": "p", - "paramLongName": "preparedInfoPath", - "paramDescription": "the path where prepared info have been stored", - "paramRequired": true - }, - { - "paramName":"test", - "paramLongName":"isTest", - "paramDescription": "true if it is executing a test", - "paramRequired": false - } - -] \ No newline at end of file diff --git a/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/resulttocommunityfromsemrel/input_preparecommunitytoresult2_parameters.json b/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/resulttocommunityfromsemrel/input_preparecommunitytoresult2_parameters.json deleted file mode 100644 index 3ba3c8e9c7..0000000000 --- a/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/resulttocommunityfromsemrel/input_preparecommunitytoresult2_parameters.json +++ /dev/null @@ -1,20 +0,0 @@ -[ - { - "paramName":"s", - "paramLongName":"sourcePath", - "paramDescription": "the path of the sequencial file to read", - "paramRequired": true - }, - { - "paramName": "ssm", - "paramLongName": "isSparkSessionManaged", - "paramDescription": "true if the spark session is managed, false otherwise", - "paramRequired": false - }, - { - "paramName": "out", - "paramLongName": "outputPath", - "paramDescription": "the path used to store temporary output files", - "paramRequired": true - } -] \ No newline at end of file diff --git a/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/resulttocommunityfromsemrel/input_preparecommunitytoresult_parameters.json b/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/resulttocommunityfromsemrel/input_preparecommunitytoresult_parameters.json deleted file mode 100644 index 271db10bb7..0000000000 --- a/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/resulttocommunityfromsemrel/input_preparecommunitytoresult_parameters.json +++ /dev/null @@ -1,44 +0,0 @@ -[ - { - "paramName":"bu", - "paramLongName":"baseURL", - "paramDescription": "URL of the isLookUp Service", - "paramRequired": true - }, - { - "paramName":"s", - "paramLongName":"sourcePath", - "paramDescription": "the path of the sequencial file to read", - "paramRequired": true - }, - { - "paramName":"as", - "paramLongName":"allowedsemrels", - "paramDescription": "the allowed semantic relations for propagation", - "paramRequired": true - }, - { - "paramName":"h", - "paramLongName":"hive_metastore_uris", - "paramDescription": "the hive metastore uris", - "paramRequired": true - }, - { - "paramName": "ssm", - "paramLongName": "isSparkSessionManaged", - "paramDescription": "true if the spark session is managed, false otherwise", - "paramRequired": false - }, - { - "paramName": "out", - "paramLongName": "outputPath", - "paramDescription": "the path used to store temporary output files", - "paramRequired": true - }, - { - "paramName":"tn", - "paramLongName":"resultTableName", - "paramDescription": "the name of the result table we are currently working on", - "paramRequired": true - } -] \ No newline at end of file diff --git a/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/resulttocommunityfromsemrel/oozie_app/config-default.xml b/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/resulttocommunityfromsemrel/oozie_app/config-default.xml deleted file mode 100644 index 2744ea92ba..0000000000 --- a/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/resulttocommunityfromsemrel/oozie_app/config-default.xml +++ /dev/null @@ -1,58 +0,0 @@ - - - jobTracker - yarnRM - - - nameNode - hdfs://nameservice1 - - - oozie.use.system.libpath - true - - - oozie.action.sharelib.for.spark - spark2 - - - hive_metastore_uris - thrift://iis-cdh5-test-m3.ocean.icm.edu.pl:9083 - - - spark2YarnHistoryServerAddress - http://iis-cdh5-test-gw.ocean.icm.edu.pl:18089 - - - spark2EventLogDir - /user/spark/spark2ApplicationHistory - - - spark2ExtraListeners - com.cloudera.spark.lineage.NavigatorAppListener - - - spark2SqlQueryExecutionListeners - com.cloudera.spark.lineage.NavigatorQueryListener - - - sparkExecutorNumber - 4 - - - sparkDriverMemory - 15G - - - sparkExecutorMemory - 6G - - - sparkExecutorCores - 1 - - - spark2MaxExecutors - 50 - - \ No newline at end of file diff --git a/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/resulttocommunityfromsemrel/oozie_app/workflow.xml b/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/resulttocommunityfromsemrel/oozie_app/workflow.xml deleted file mode 100644 index 916eb8b7ce..0000000000 --- a/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/resulttocommunityfromsemrel/oozie_app/workflow.xml +++ /dev/null @@ -1,366 +0,0 @@ - - - - sourcePath - the source path - - - allowedsemrels - the semantic relationships allowed for propagation - - - baseURL - the baseurl for the comminity APIs - - - outputPath - the output path - - - - - - - Action failed, error message[${wf:errorMessage(wf:lastErrorNode())}] - - - - - - - - - - - - - - - - - - - - - ${jobTracker} - ${nameNode} - ${nameNode}/${sourcePath}/relation - ${nameNode}/${outputPath}/relation - - - - - - - - ${jobTracker} - ${nameNode} - ${nameNode}/${sourcePath}/organization - ${nameNode}/${outputPath}/organization - - - - - - - - ${jobTracker} - ${nameNode} - ${nameNode}/${sourcePath}/project - ${nameNode}/${outputPath}/project - - - - - - - - ${jobTracker} - ${nameNode} - ${nameNode}/${sourcePath}/datasource - ${nameNode}/${outputPath}/datasource - - - - - - - - - - - - - - - - - yarn - cluster - ResultToCommunitySemRel-PreparePhase1-Publications - eu.dnetlib.dhp.resulttocommunityfromsemrel.PrepareResultCommunitySetStep1 - dhp-enrichment-${projectVersion}.jar - - --executor-cores=${sparkExecutorCores} - --executor-memory=${sparkExecutorMemory} - --driver-memory=${sparkDriverMemory} - --conf spark.extraListeners=${spark2ExtraListeners} - --conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners} - --conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress} - --conf spark.eventLog.dir=${nameNode}${spark2EventLogDir} - --conf spark.dynamicAllocation.enabled=true - --conf spark.dynamicAllocation.maxExecutors=${spark2MaxExecutors} - - --sourcePath${sourcePath} - --hive_metastore_uris${hive_metastore_uris} - --resultTableNameeu.dnetlib.dhp.schema.oaf.Publication - --outputPath${workingDir}/preparedInfo/targetCommunityAssoc - --allowedsemrels${allowedsemrels} - --baseURL${baseURL} - - - - - - - - yarn - cluster - ResultToCommunitySemRel-PreparePhase1-Dataset - eu.dnetlib.dhp.resulttocommunityfromsemrel.PrepareResultCommunitySetStep1 - dhp-enrichment-${projectVersion}.jar - - --executor-cores=${sparkExecutorCores} - --executor-memory=${sparkExecutorMemory} - --driver-memory=${sparkDriverMemory} - --conf spark.extraListeners=${spark2ExtraListeners} - --conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners} - --conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress} - --conf spark.eventLog.dir=${nameNode}${spark2EventLogDir} - --conf spark.dynamicAllocation.enabled=true - --conf spark.dynamicAllocation.maxExecutors=${spark2MaxExecutors} - - --sourcePath${sourcePath} - --hive_metastore_uris${hive_metastore_uris} - --resultTableNameeu.dnetlib.dhp.schema.oaf.Dataset - --outputPath${workingDir}/preparedInfo/targetCommunityAssoc - --allowedsemrels${allowedsemrels} - --isLookUpUrl${isLookUpUrl} - - - - - - - - yarn - cluster - ResultToCommunitySemRel-PreparePhase1-ORP - eu.dnetlib.dhp.resulttocommunityfromsemrel.PrepareResultCommunitySetStep1 - dhp-enrichment-${projectVersion}.jar - - --executor-cores=${sparkExecutorCores} - --executor-memory=${sparkExecutorMemory} - --driver-memory=${sparkDriverMemory} - --conf spark.extraListeners=${spark2ExtraListeners} - --conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners} - --conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress} - --conf spark.eventLog.dir=${nameNode}${spark2EventLogDir} - --conf spark.dynamicAllocation.enabled=true - --conf spark.dynamicAllocation.maxExecutors=${spark2MaxExecutors} - - --sourcePath${sourcePath} - --hive_metastore_uris${hive_metastore_uris} - --resultTableNameeu.dnetlib.dhp.schema.oaf.OtherResearchProduct - --outputPath${workingDir}/preparedInfo/targetCommunityAssoc - --allowedsemrels${allowedsemrels} - --isLookUpUrl${isLookUpUrl} - - - - - - - - yarn - cluster - ResultToCommunitySemRel-PreparePhase1-Software - eu.dnetlib.dhp.resulttocommunityfromsemrel.PrepareResultCommunitySetStep1 - dhp-enrichment-${projectVersion}.jar - - --executor-cores=${sparkExecutorCores} - --executor-memory=${sparkExecutorMemory} - --driver-memory=${sparkDriverMemory} - --conf spark.extraListeners=${spark2ExtraListeners} - --conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners} - --conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress} - --conf spark.eventLog.dir=${nameNode}${spark2EventLogDir} - --conf spark.dynamicAllocation.enabled=true - --conf spark.dynamicAllocation.maxExecutors=${spark2MaxExecutors} - - --sourcePath${sourcePath} - --hive_metastore_uris${hive_metastore_uris} - --resultTableNameeu.dnetlib.dhp.schema.oaf.Software - --outputPath${workingDir}/preparedInfo/targetCommunityAssoc - --allowedsemrels${allowedsemrels} - --isLookUpUrl${isLookUpUrl} - - - - - - - - - - yarn - cluster - ResultToCommunityEmRelPropagation-PreparePhase2 - eu.dnetlib.dhp.resulttocommunityfromsemrel.PrepareResultCommunitySetStep2 - dhp-enrichment-${projectVersion}.jar - - --executor-cores=${sparkExecutorCores} - --executor-memory=${sparkExecutorMemory} - --driver-memory=${sparkDriverMemory} - --conf spark.extraListeners=${spark2ExtraListeners} - --conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners} - --conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress} - --conf spark.eventLog.dir=${nameNode}${spark2EventLogDir} - --conf spark.dynamicAllocation.enabled=true - --conf spark.dynamicAllocation.maxExecutors=${spark2MaxExecutors} - - --sourcePath${workingDir}/preparedInfo/targetCommunityAssoc - --outputPath${workingDir}/preparedInfo/mergedCommunityAssoc - - - - - - - - - - - - - - - yarn - cluster - Result2CommunitySemRelPropagation-Publication - eu.dnetlib.dhp.resulttocommunityfromsemrel.SparkResultToCommunityThroughSemRelJob - dhp-enrichment-${projectVersion}.jar - - --executor-cores=${sparkExecutorCores} - --executor-memory=${sparkExecutorMemory} - --driver-memory=${sparkDriverMemory} - --conf spark.extraListeners=${spark2ExtraListeners} - --conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners} - --conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress} - --conf spark.eventLog.dir=${nameNode}${spark2EventLogDir} - --conf spark.dynamicAllocation.enabled=true - --conf spark.dynamicAllocation.maxExecutors=${spark2MaxExecutors} - - --preparedInfoPath${workingDir}/preparedInfo/mergedCommunityAssoc - --sourcePath${sourcePath}/publication - --hive_metastore_uris${hive_metastore_uris} - --resultTableNameeu.dnetlib.dhp.schema.oaf.Publication - --outputPath${outputPath}/publication - --saveGraph${saveGraph} - - - - - - - - yarn - cluster - Result2CommunitySemRelPropagation-Dataset - eu.dnetlib.dhp.resulttocommunityfromsemrel.SparkResultToCommunityThroughSemRelJob - dhp-enrichment-${projectVersion}.jar - - --executor-cores=${sparkExecutorCores} - --executor-memory=${sparkExecutorMemory} - --driver-memory=${sparkDriverMemory} - --conf spark.extraListeners=${spark2ExtraListeners} - --conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners} - --conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress} - --conf spark.eventLog.dir=${nameNode}${spark2EventLogDir} - --conf spark.dynamicAllocation.enabled=true - --conf spark.dynamicAllocation.maxExecutors=${spark2MaxExecutors} - - --preparedInfoPath${workingDir}/preparedInfo/mergedCommunityAssoc - --sourcePath${sourcePath}/dataset - --hive_metastore_uris${hive_metastore_uris} - --resultTableNameeu.dnetlib.dhp.schema.oaf.Dataset - --outputPath${outputPath}/dataset - --saveGraph${saveGraph} - - - - - - - - yarn - cluster - Result2CommunitySemRelPropagation-ORP - eu.dnetlib.dhp.resulttocommunityfromsemrel.SparkResultToCommunityThroughSemRelJob - dhp-enrichment-${projectVersion}.jar - - --executor-cores=${sparkExecutorCores} - --executor-memory=${sparkExecutorMemory} - --driver-memory=${sparkDriverMemory} - --conf spark.extraListeners=${spark2ExtraListeners} - --conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners} - --conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress} - --conf spark.eventLog.dir=${nameNode}${spark2EventLogDir} - --conf spark.dynamicAllocation.enabled=true - --conf spark.dynamicAllocation.maxExecutors=${spark2MaxExecutors} - - --preparedInfoPath${workingDir}/preparedInfo/mergedCommunityAssoc - --sourcePath${sourcePath}/otherresearchproduct - --hive_metastore_uris${hive_metastore_uris} - --resultTableNameeu.dnetlib.dhp.schema.oaf.OtherResearchProduct - --outputPath${outputPath}/otherresearchproduct - --saveGraph${saveGraph} - - - - - - - - yarn - cluster - Result2CommunitySemRelPropagation-Software - eu.dnetlib.dhp.resulttocommunityfromsemrel.SparkResultToCommunityThroughSemRelJob - dhp-enrichment-${projectVersion}.jar - - --executor-cores=${sparkExecutorCores} - --executor-memory=${sparkExecutorMemory} - --driver-memory=${sparkDriverMemory} - --conf spark.extraListeners=${spark2ExtraListeners} - --conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners} - --conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress} - --conf spark.eventLog.dir=${nameNode}${spark2EventLogDir} - --conf spark.dynamicAllocation.enabled=true - --conf spark.dynamicAllocation.maxExecutors=${spark2MaxExecutors} - - --preparedInfoPath${workingDir}/preparedInfo/mergedCommunityAssoc - --sourcePath${sourcePath}/software - --hive_metastore_uris${hive_metastore_uris} - --resultTableNameeu.dnetlib.dhp.schema.oaf.Software - --outputPath${outputPath}/software - --saveGraph${saveGraph} - - - - - - - - - - \ No newline at end of file diff --git a/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/resulttoorganizationfrominstrepo/input_newrelation_parameters.json b/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/resulttoorganizationfrominstrepo/input_newrelation_parameters.json deleted file mode 100644 index 5fe92cff13..0000000000 --- a/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/resulttoorganizationfrominstrepo/input_newrelation_parameters.json +++ /dev/null @@ -1,20 +0,0 @@ -[ - { - "paramName":"s", - "paramLongName":"sourcePath", - "paramDescription": "the path of the sequencial file to read", - "paramRequired": true - }, - - { - "paramName": "ssm", - "paramLongName": "isSparkSessionManaged", - "paramDescription": "the path where prepared info have been stored", - "paramRequired": false - },{ - "paramName": "o", - "paramLongName": "outputPath", - "paramDescription": "institutional repositories that should not be considered for the propagation", - "paramRequired": false -} -] \ No newline at end of file diff --git a/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/resulttoorganizationfrominstrepo/input_prepareresultorg_parameters.json b/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/resulttoorganizationfrominstrepo/input_prepareresultorg_parameters.json deleted file mode 100644 index 3f4b1d151b..0000000000 --- a/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/resulttoorganizationfrominstrepo/input_prepareresultorg_parameters.json +++ /dev/null @@ -1,32 +0,0 @@ -[ - { - "paramName":"s", - "paramLongName":"sourcePath", - "paramDescription": "the path of the sequencial file to read", - "paramRequired": true - }, - { - "paramName":"h", - "paramLongName":"hive_metastore_uris", - "paramDescription": "the hive metastore uris", - "paramRequired": true - }, - - { - "paramName":"wp", - "paramLongName":"workingPath", - "paramDescription": "the working path", - "paramRequired": true - }, - { - "paramName": "ssm", - "paramLongName": "isSparkSessionManaged", - "paramDescription": "the path where prepared info have been stored", - "paramRequired": false - },{ - "paramName": "bl", - "paramLongName": "blacklist", - "paramDescription": "institutional repositories that should not be considered for the propagation", - "paramRequired": false -} -] \ No newline at end of file diff --git a/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/resulttoorganizationfrominstrepo/input_propagationresulaffiliationfrominstrepo_parameters.json b/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/resulttoorganizationfrominstrepo/input_propagationresulaffiliationfrominstrepo_parameters.json deleted file mode 100644 index d2b076c827..0000000000 --- a/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/resulttoorganizationfrominstrepo/input_propagationresulaffiliationfrominstrepo_parameters.json +++ /dev/null @@ -1,56 +0,0 @@ -[ - { - "paramName":"s", - "paramLongName":"sourcePath", - "paramDescription": "the path of the sequencial file to read", - "paramRequired": true - }, - { - "paramName":"h", - "paramLongName":"hive_metastore_uris", - "paramDescription": "the hive metastore uris", - "paramRequired": true - }, - { - "paramName":"sg", - "paramLongName":"saveGraph", - "paramDescription": "true if the new version of the graph must be saved", - "paramRequired": false - }, - { - "paramName":"dop", - "paramLongName":"datasourceOrganizationPath", - "paramDescription": "path where to store/find association from datasource and organization", - "paramRequired": true - }, - { - "paramName":"alp", - "paramLongName":"alreadyLinkedPath", - "paramDescription": "path where to store/find already linked results and organizations", - "paramRequired": true - }, - { - "paramName": "ssm", - "paramLongName": "isSparkSessionManaged", - "paramDescription": "the path where prepared info have been stored", - "paramRequired": false - }, - { - "paramName": "test", - "paramLongName": "isTest", - "paramDescription": "true if it is a test running", - "paramRequired": false - }, - { - "paramName":"tn", - "paramLongName":"resultTableName", - "paramDescription": "the name of the result table we are currently working on", - "paramRequired": true - }, - { - "paramName": "out", - "paramLongName": "outputPath", - "paramDescription": "the path used to store temporary output files", - "paramRequired": true - } -] \ No newline at end of file diff --git a/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/resulttoorganizationfrominstrepo/oozie_app/config-default.xml b/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/resulttoorganizationfrominstrepo/oozie_app/config-default.xml deleted file mode 100644 index 2744ea92ba..0000000000 --- a/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/resulttoorganizationfrominstrepo/oozie_app/config-default.xml +++ /dev/null @@ -1,58 +0,0 @@ - - - jobTracker - yarnRM - - - nameNode - hdfs://nameservice1 - - - oozie.use.system.libpath - true - - - oozie.action.sharelib.for.spark - spark2 - - - hive_metastore_uris - thrift://iis-cdh5-test-m3.ocean.icm.edu.pl:9083 - - - spark2YarnHistoryServerAddress - http://iis-cdh5-test-gw.ocean.icm.edu.pl:18089 - - - spark2EventLogDir - /user/spark/spark2ApplicationHistory - - - spark2ExtraListeners - com.cloudera.spark.lineage.NavigatorAppListener - - - spark2SqlQueryExecutionListeners - com.cloudera.spark.lineage.NavigatorQueryListener - - - sparkExecutorNumber - 4 - - - sparkDriverMemory - 15G - - - sparkExecutorMemory - 6G - - - sparkExecutorCores - 1 - - - spark2MaxExecutors - 50 - - \ No newline at end of file diff --git a/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/resulttoorganizationfrominstrepo/oozie_app/workflow.xml b/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/resulttoorganizationfrominstrepo/oozie_app/workflow.xml deleted file mode 100644 index edfff8817d..0000000000 --- a/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/resulttoorganizationfrominstrepo/oozie_app/workflow.xml +++ /dev/null @@ -1,277 +0,0 @@ - - - - sourcePath - the source path - - - outputPath - sets the outputPath - - - - - ${jobTracker} - ${nameNode} - - - oozie.action.sharelib.for.spark - ${oozieActionShareLibForSpark2} - - - - - - - - Action failed, error message[${wf:errorMessage(wf:lastErrorNode())}] - - - - - - - - - - - - - - - - - - - - - - - - - ${nameNode}/${sourcePath}/relation - ${nameNode}/${outputPath}/relation - - - - - - - - ${nameNode}/${sourcePath}/publication - ${nameNode}/${outputPath}/publication - - - - - - - - ${nameNode}/${sourcePath}/dataset - ${nameNode}/${outputPath}/dataset - - - - - - - - ${nameNode}/${sourcePath}/otherresearchproduct - ${nameNode}/${outputPath}/otherresearchproduct - - - - - - - - ${nameNode}/${sourcePath}/software - ${nameNode}/${outputPath}/software - - - - - - - - ${nameNode}/${sourcePath}/organization - ${nameNode}/${outputPath}/organization - - - - - - - - ${nameNode}/${sourcePath}/project - ${nameNode}/${outputPath}/project - - - - - - - - ${nameNode}/${sourcePath}/datasource - ${nameNode}/${outputPath}/datasource - - - - - - - - - - yarn - cluster - PrepareResultOrganizationAssociation - eu.dnetlib.dhp.resulttoorganizationfrominstrepo.PrepareResultInstRepoAssociation - dhp-enrichment-${projectVersion}.jar - - --executor-cores=${sparkExecutorCores} - --executor-memory=${sparkExecutorMemory} - --driver-memory=${sparkDriverMemory} - --conf spark.extraListeners=${spark2ExtraListeners} - --conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners} - --conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress} - --conf spark.eventLog.dir=${nameNode}${spark2EventLogDir} - - --sourcePath${sourcePath} - --hive_metastore_uris${hive_metastore_uris} - --datasourceOrganizationPath${workingDir}/preparedInfo/datasourceOrganization - --alreadyLinkedPath${workingDir}/preparedInfo/alreadyLinked - --blacklist${blacklist} - - - - - - - - - - - - - - - yarn - cluster - resultToOrganizationFromInstRepoPropagationForPublications - eu.dnetlib.dhp.resulttoorganizationfrominstrepo.SparkResultToOrganizationFromIstRepoJob - dhp-enrichment-${projectVersion}.jar - - --executor-cores=${sparkExecutorCores} - --executor-memory=${sparkExecutorMemory} - --driver-memory=${sparkDriverMemory} - --conf spark.extraListeners=${spark2ExtraListeners} - --conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners} - --conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress} - --conf spark.eventLog.dir=${nameNode}${spark2EventLogDir} - --conf spark.dynamicAllocation.enabled=true - --conf spark.dynamicAllocation.maxExecutors=${spark2MaxExecutors} - - --sourcePath${sourcePath}/publication - --outputPath${outputPath}/relation - --datasourceOrganizationPath${workingDir}/preparedInfo/datasourceOrganization - --alreadyLinkedPath${workingDir}/preparedInfo/alreadyLinked - --hive_metastore_uris${hive_metastore_uris} - --resultTableNameeu.dnetlib.dhp.schema.oaf.Publication - - - - - - - - yarn - cluster - resultToOrganizationFromInstRepoPropagationForDataset - eu.dnetlib.dhp.resulttoorganizationfrominstrepo.SparkResultToOrganizationFromIstRepoJob - dhp-enrichment-${projectVersion}.jar - - --executor-cores=${sparkExecutorCores} - --executor-memory=${sparkExecutorMemory} - --driver-memory=${sparkDriverMemory} - --conf spark.extraListeners=${spark2ExtraListeners} - --conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners} - --conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress} - --conf spark.eventLog.dir=${nameNode}${spark2EventLogDir} - --conf spark.dynamicAllocation.enabled=true - --conf spark.dynamicAllocation.maxExecutors=${spark2MaxExecutors} - - --sourcePath${sourcePath}/dataset - --outputPath${outputPath}/relation - --datasourceOrganizationPath${workingDir}/preparedInfo/datasourceOrganization - --alreadyLinkedPath${workingDir}/preparedInfo/alreadyLinked - --hive_metastore_uris${hive_metastore_uris} - --resultTableNameeu.dnetlib.dhp.schema.oaf.Dataset - - - - - - - - yarn - cluster - resultToOrganizationFromInstRepoPropagationForORP - eu.dnetlib.dhp.resulttoorganizationfrominstrepo.SparkResultToOrganizationFromIstRepoJob - dhp-enrichment-${projectVersion}.jar - - --executor-cores=${sparkExecutorCores} - --executor-memory=${sparkExecutorMemory} - --driver-memory=${sparkDriverMemory} - --conf spark.extraListeners=${spark2ExtraListeners} - --conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners} - --conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress} - --conf spark.eventLog.dir=${nameNode}${spark2EventLogDir} - --conf spark.dynamicAllocation.enabled=true - --conf spark.dynamicAllocation.maxExecutors=${spark2MaxExecutors} - - --sourcePath${sourcePath}/otherresearchproduct - --outputPath${outputPath}/relation - --datasourceOrganizationPath${workingDir}/preparedInfo/datasourceOrganization - --alreadyLinkedPath${workingDir}/preparedInfo/alreadyLinked - --hive_metastore_uris${hive_metastore_uris} - --resultTableNameeu.dnetlib.dhp.schema.oaf.OtherResearchProduct - - - - - - - - yarn - cluster - resultToOrganizationFromInstRepoPropagationForSoftware - eu.dnetlib.dhp.resulttoorganizationfrominstrepo.SparkResultToOrganizationFromIstRepoJob - dhp-enrichment-${projectVersion}.jar - - --executor-cores=${sparkExecutorCores} - --executor-memory=${sparkExecutorMemory} - --driver-memory=${sparkDriverMemory} - --conf spark.extraListeners=${spark2ExtraListeners} - --conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners} - --conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress} - --conf spark.eventLog.dir=${nameNode}${spark2EventLogDir} - --conf spark.dynamicAllocation.enabled=true - --conf spark.dynamicAllocation.maxExecutors=${spark2MaxExecutors} - - --sourcePath${sourcePath}/software - --outputPath${outputPath}/relation - --datasourceOrganizationPath${workingDir}/preparedInfo/datasourceOrganization - --alreadyLinkedPath${workingDir}/preparedInfo/alreadyLinked - --hive_metastore_uris${hive_metastore_uris} - --resultTableNameeu.dnetlib.dhp.schema.oaf.Software - - - - - - - - - - \ No newline at end of file diff --git a/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/wf/main/job.properties b/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/wf/main/job.properties index 93e9e0ab1d..4cb759343c 100644 --- a/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/wf/main/job.properties +++ b/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/wf/main/job.properties @@ -1,12 +1,12 @@ sourcePath=/tmp/beta_provision/graph/09_graph_dedup_enriched -resumeFrom=AffiliationSemanticRelation +resumeFrom=CountryPropagation allowedsemrelsorcidprop=isSupplementedBy;isSupplementTo allowedsemrelsresultproject=isSupplementedBy;isSupplementTo allowedsemrelscommunitysemrel=isSupplementedBy;isSupplementTo -datasourceWhitelistForCountryPropagation=10|openaire____::3795d6478e30e2c9f787d427ff160944;10|opendoar____::16e6a3326dd7d868cbc926602a61e4d0;10|eurocrisdris::fe4903425d9040f680d8610d9079ea14 -allowedtypes=pubsrepository::institutional +datasourceWhitelistForCountryPropagation=10|opendoar____::16e6a3326dd7d868cbc926602a61e4d0;10|openaire____::fdb035c8b3e0540a8d9a561a6c44f4de;10|eurocrisdris::fe4903425d9040f680d8610d9079ea14;10|openaire____::5b76240cc27a58c6f7ceef7d8c36660e;10|openaire____::172bbccecf8fca44ab6a6653e84cb92a;10|openaire____::149c6590f8a06b46314eed77bfca693f;10|eurocrisdris::a6026877c1a174d60f81fd71f62df1c1;10|openaire____::4692342f0992d91f9e705c26959f09e0;10|openaire____::8d529dbb05ec0284662b391789e8ae2a;10|openaire____::345c9d171ef3c5d706d08041d506428c;10|opendoar____::1c1d4df596d01da60385f0bb17a4a9e0;10|opendoar____::7a614fd06c325499f1680b9896beedeb;10|opendoar____::1ee3dfcd8a0645a25a35977997223d22;10|opendoar____::d296c101daa88a51f6ca8cfc1ac79b50;10|opendoar____::798ed7d4ee7138d49b8828958048130a;10|openaire____::c9d2209ecc4d45ba7b4ca7597acb88a2;10|eurocrisdris::c49e0fe4b9ba7b7fab717d1f0f0a674d;10|eurocrisdris::9ae43d14471c4b33661fedda6f06b539;10|eurocrisdris::432ca599953ff50cd4eeffe22faf3e48 +#allowedtypes=pubsrepository::institutional +allowedtypes=Institutional outputPath=/tmp/miriam/enrichment_one_step -organizationtoresultcommunitymap={"20|corda__h2020::3fb05a9524c3f790391261347852f638":["mes","euromarine"], "20|corda__h2020::e8dbe14cca9bf6fce09d468872f813f8":["mes","euromarine"], "20|snsf________::9b253f265e3bef5cae6d881fdf61aceb":["mes","euromarine"],"20|ukri________::e054eea0a47665af8c3656b5785ccf76":["mes","euromarine"],"20|corda__h2020::edc18d67c9b11fb616ca9f6e1db1b151":["mes","euromarine"],"20|ukri________::d5736d9da90521ddcdc7828a05a85e9a":["mes","euromarine"],"20|corda__h2020::f5d418d3aa1cf817ddefcc3fdc039f27":["mes","euromarine"],"20|snsf________::8fa091f8f25a846779acb4ea97b50aef":["mes","euromarine"],"20|corda__h2020::81e020977211c2c40fae2e1a50bffd71":["mes","euromarine"],"20|corda_______::81e020977211c2c40fae2e1a50bffd71":["mes","euromarine"],"20|snsf________::31d0a100e54e3cdb3c6f52d91e638c78":["mes","euromarine"],"20|corda__h2020::ea379ef91b8cc86f9ac5edc4169292db":["mes","euromarine"],"20|corda__h2020::f75ee2ee48e5cb0ec8c8d30aaa8fef70":["mes","euromarine"],"20|ukri________::e16010089551a1a9182a94604fc0ea59":["mes","euromarine"],"20|corda__h2020::38531a2cce7c5c347ffc439b07c1f43b":["mes","euromarine"],"20|corda_______::38531a2cce7c5c347ffc439b07c1f43b":["mes","euromarine"],"20|grid________::b2cbbf5eadbbf87d534b022bad3191d7":["mes","euromarine"],"20|snsf________::74730ef1439d7f7636a8be58a6b471b8":["mes","euromarine"],"20|nsf_________::ad72e19043a5a467e35f9b444d11563e":["mes","euromarine"],"20|ukri________::0fc3e92500290902a2d38ec2445e74c3":["mes","euromarine"],"20|grid________::ad2c29905da0eb3c06b3fa80cacd89ea":["mes","euromarine"],"20|corda__h2020::30b53e4d63d3724f00acb9cbaca40860":["mes","euromarine"],"20|corda__h2020::f60f84bee14ad93f0db0e49af1d5c317":["mes","euromarine"], "20|corda__h2020::7bf251ac3765b5e89d82270a1763d09f":["mes","euromarine"], "20|corda__h2020::65531bd11be9935948c7f2f4db1c1832":["mes","euromarine"], "20|corda__h2020::e0e98f86bbc76638bbb72a8fe2302946":["mes","euromarine"], "20|snsf________::3eb43582ac27601459a8d8b3e195724b":["mes","euromarine"], "20|corda__h2020::af2481dab65d06c8ea0ae02b5517b9b6":["mes","euromarine"], "20|corda__h2020::c19d05cfde69a50d3ebc89bd0ee49929":["mes","euromarine"], "20|corda__h2020::af0bfd9fc09f80d9488f56d71a9832f0":["mes","euromarine"], "20|ukri________::f33c02afb0dc66c49d0ed97ca5dd5cb0":["beopen"], "20|grid________::a867f78acdc5041b34acfe4f9a349157":["beopen"], "20|grid________::7bb116a1a9f95ab812bf9d2dea2be1ff":["beopen"], "20|corda__h2020::6ab0e0739dbe625b99a2ae45842164ad":["beopen"], "20|corda__h2020::8ba50792bc5f4d51d79fca47d860c602":["beopen"], "20|corda_______::8ba50792bc5f4d51d79fca47d860c602":["beopen"], "20|corda__h2020::e70e9114979e963eef24666657b807c3":["beopen"], "20|corda_______::e70e9114979e963eef24666657b807c3":["beopen"], "20|corda_______::15911e01e9744d57205825d77c218737":["beopen"], "20|opendoar____::056a41e24e2a9a67215e87bbee6a80ab":["beopen"], "20|opendoar____::7f67f2e6c6fbb0628f8160fcd3d92ae3":["beopen"], "20|grid________::a8ecfd7c084e561168bcbe6bf0daf3e3":["beopen"], "20|corda_______::7bbe6cc5d8ec1864739a04b0d020c9e9":["beopen"], "20|corda_______::3ff558e30c2e434d688539548300b050":["beopen"], "20|corda__h2020::5ffee5b3b83b33a8cf0e046877bd3a39":["beopen"], "20|corda__h2020::5187217e2e806a6df3579c46f82401bc":["beopen"], "20|grid________::5fa7e2709bcd945e26bfa18689adeec1":["beopen"], "20|corda_______::d8696683c53027438031a96ad27c3c07":["beopen"], "20|corda__h2020::d8696683c53027438031a96ad27c3c07":["beopen"], "20|ukri________::23a79ebdfa59790864e4a485881568c1":["beopen"], "20|corda__h2020::b76cf8fe49590a966953c37e18608af9":["beopen"], "20|grid________::d2f0204126ee709244a488a4cd3b91c2":["beopen"], "20|corda__h2020::05aba9d2ed17533d15221e5655ac11e6":["beopen"], "20|grid________::802401579481dc32062bdee69f5e6a34":["beopen"], "20|corda__h2020::3f6d9d54cac975a517ba6b252c81582d":["beopen"], "20|openorgs____::d11f981828c485cd23d93f7f24f24db1":["eut"], "20|openorgs____::e66fe5dd092752e1dd6fd29fc699933a":["eut"], "20|openorgs____::526468206bca24c1c90da6a312295cf4":["eut"], "20|openorgs____::08e311e656e65ccb32e07c66b15b6ff7":["eut"], "20|openorgs____::55a1f889758964b77682904218fdb298":["eut"], "20|openorgs____::530092b6970d60a5329beb9f39e8d7d4":["eut"], "20|openorgs____::aadafa39392b3e200102596a3a4aad9d":["eut"], "20|openorgs____::c3fe999c74fad308132b8a5971367dce":["eut"], "20|openorgs____::1624ff7c01bb641b91f4518539a0c28a":["aurora"], "20|openorgs____::cdda7cfe17c89eb50628ec2eb1f8acd2":["aurora"], "20|openorgs____::818b75030e0e40612d69e049843ede7e":["aurora"], "20|openorgs____::0b0102bae51f4f4ef5ba57fbe1523b92":["aurora"], "20|openorgs____::ed47496b44722f0e9d7b98898189be0d":["aurora"], "20|openorgs____::eb0669daa9efeb898a3090d8aac7c953":["aurora"], "20|openorgs____::eb391317ed0dc684aa81ac16265de041":["aurora"], "20|openorgs____::f7cfcc98245e22c7d6e321cde930e746":["aurora"], "20|openorgs____::f33179d3306ba2599f7a898b056b604f":["aurora"], "20|pending_org_::75c41e6dd18466709ef359323d96fa05":["aurora"]} pathMap ={"author":"$['author'][*]['fullname']", \ "title":"$['title'][*]['value']",\ "orcid":"$['author'][*]['pid'][*][?(@['qualifier']['classid']=='orcid')]['value']" ,\ diff --git a/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/wf/main/oozie_app/workflow.xml b/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/wf/main/oozie_app/workflow.xml index de054b962b..8e91707b6e 100644 --- a/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/wf/main/oozie_app/workflow.xml +++ b/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/wf/main/oozie_app/workflow.xml @@ -29,10 +29,6 @@ outputPath the output path - - organizationtoresultcommunitymap - organization community map - pathMap the json path associated to each selection field @@ -315,7 +311,7 @@ allowedtypes - ${allowedtupes} + ${allowedtypes} diff --git a/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/wf/subworkflows/bulktag/oozie_app/workflow.xml b/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/wf/subworkflows/bulktag/oozie_app/workflow.xml index 307997d4cd..6c51634484 100644 --- a/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/wf/subworkflows/bulktag/oozie_app/workflow.xml +++ b/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/wf/subworkflows/bulktag/oozie_app/workflow.xml @@ -26,12 +26,20 @@ - + Action failed, error message[${wf:errorMessage(wf:lastErrorNode())}] + + + + + + + + diff --git a/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/wf/subworkflows/countrypropagation/oozie_app/workflow.xml b/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/wf/subworkflows/countrypropagation/oozie_app/workflow.xml index 1fbaeb5d55..933bab7e06 100644 --- a/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/wf/subworkflows/countrypropagation/oozie_app/workflow.xml +++ b/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/wf/subworkflows/countrypropagation/oozie_app/workflow.xml @@ -25,12 +25,20 @@ - + Action failed, error message[${wf:errorMessage(wf:lastErrorNode())}] + + + + + + + + diff --git a/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/wf/subworkflows/entitytoorganizationfromsemrel/oozie_app/workflow.xml b/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/wf/subworkflows/entitytoorganizationfromsemrel/oozie_app/workflow.xml index dbb22b9948..05824d209b 100644 --- a/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/wf/subworkflows/entitytoorganizationfromsemrel/oozie_app/workflow.xml +++ b/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/wf/subworkflows/entitytoorganizationfromsemrel/oozie_app/workflow.xml @@ -18,12 +18,20 @@ - + Action failed, error message[${wf:errorMessage(wf:lastErrorNode())}] + + + + + + + + diff --git a/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/wf/subworkflows/projecttoresult/oozie_app/workflow.xml b/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/wf/subworkflows/projecttoresult/oozie_app/workflow.xml index 93a2f98be3..f0db9c777f 100644 --- a/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/wf/subworkflows/projecttoresult/oozie_app/workflow.xml +++ b/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/wf/subworkflows/projecttoresult/oozie_app/workflow.xml @@ -22,13 +22,21 @@ - + Action failed, error message[${wf:errorMessage(wf:lastErrorNode())}] - + + + + + + + + + yarn diff --git a/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/wf/subworkflows/resulttocommunityfromorganization/oozie_app/workflow.xml b/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/wf/subworkflows/resulttocommunityfromorganization/oozie_app/workflow.xml index 8aec530cc9..6aeffb4574 100644 --- a/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/wf/subworkflows/resulttocommunityfromorganization/oozie_app/workflow.xml +++ b/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/wf/subworkflows/resulttocommunityfromorganization/oozie_app/workflow.xml @@ -21,12 +21,21 @@ - + Action failed, error message[${wf:errorMessage(wf:lastErrorNode())}] + + + + + + + + + yarn @@ -75,9 +84,9 @@ --conf spark.eventLog.dir=${nameNode}${spark2EventLogDir} --conf spark.dynamicAllocation.maxExecutors=${spark2MaxExecutors} - --preparedInfoPath${workingDir}/preparedInfo/resultCommunityList + --preparedInfoPath${workingDir}/communityorganization/preparedInfo/resultCommunityList --sourcePath${sourcePath}/ - --outputPath${workingDir}/resulttocommunityfromorganization/ + --outputPath${workingDir}/communityorganization/resulttocommunityfromorganization/ diff --git a/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/wf/subworkflows/resulttocommunityfromproject/oozie_app/workflow.xml b/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/wf/subworkflows/resulttocommunityfromproject/oozie_app/workflow.xml index 90ed2e0b6f..dd845064b2 100644 --- a/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/wf/subworkflows/resulttocommunityfromproject/oozie_app/workflow.xml +++ b/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/wf/subworkflows/resulttocommunityfromproject/oozie_app/workflow.xml @@ -21,12 +21,19 @@ - + Action failed, error message[${wf:errorMessage(wf:lastErrorNode())}] - + + + + + + + + diff --git a/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/wf/subworkflows/resulttocommunityfromsemrel/oozie_app/workflow.xml b/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/wf/subworkflows/resulttocommunityfromsemrel/oozie_app/workflow.xml index be88c45bdc..773c7fba76 100644 --- a/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/wf/subworkflows/resulttocommunityfromsemrel/oozie_app/workflow.xml +++ b/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/wf/subworkflows/resulttocommunityfromsemrel/oozie_app/workflow.xml @@ -18,13 +18,20 @@ - + Action failed, error message[${wf:errorMessage(wf:lastErrorNode())}] - + + + + + + + + @@ -41,8 +48,10 @@ eu.dnetlib.dhp.resulttocommunityfromsemrel.PrepareResultCommunitySetStep1 dhp-enrichment-${projectVersion}.jar - --executor-cores=${sparkExecutorCores} - --executor-memory=${sparkExecutorMemory} + --executor-cores=6 + --executor-memory=5G + --conf spark.executor.memoryOverhead=3g + --conf spark.sql.shuffle.partitions=3284 --driver-memory=${sparkDriverMemory} --conf spark.extraListeners=${spark2ExtraListeners} --conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners} @@ -70,8 +79,10 @@ eu.dnetlib.dhp.resulttocommunityfromsemrel.PrepareResultCommunitySetStep1 dhp-enrichment-${projectVersion}.jar - --executor-cores=${sparkExecutorCores} - --executor-memory=${sparkExecutorMemory} + --executor-cores=6 + --executor-memory=5G + --conf spark.executor.memoryOverhead=3g + --conf spark.sql.shuffle.partitions=3284 --driver-memory=${sparkDriverMemory} --conf spark.extraListeners=${spark2ExtraListeners} --conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners} @@ -99,8 +110,10 @@ eu.dnetlib.dhp.resulttocommunityfromsemrel.PrepareResultCommunitySetStep1 dhp-enrichment-${projectVersion}.jar - --executor-cores=${sparkExecutorCores} - --executor-memory=${sparkExecutorMemory} + --executor-cores=6 + --executor-memory=5G + --conf spark.executor.memoryOverhead=3g + --conf spark.sql.shuffle.partitions=3284 --driver-memory=${sparkDriverMemory} --conf spark.extraListeners=${spark2ExtraListeners} --conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners} @@ -128,8 +141,10 @@ eu.dnetlib.dhp.resulttocommunityfromsemrel.PrepareResultCommunitySetStep1 dhp-enrichment-${projectVersion}.jar - --executor-cores=${sparkExecutorCores} - --executor-memory=${sparkExecutorMemory} + --executor-cores=6 + --executor-memory=5G + --conf spark.executor.memoryOverhead=3g + --conf spark.sql.shuffle.partitions=3284 --driver-memory=${sparkDriverMemory} --conf spark.extraListeners=${spark2ExtraListeners} --conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners} @@ -159,8 +174,10 @@ eu.dnetlib.dhp.resulttocommunityfromsemrel.PrepareResultCommunitySetStep2 dhp-enrichment-${projectVersion}.jar - --executor-cores=${sparkExecutorCores} - --executor-memory=${sparkExecutorMemory} + --executor-cores=6 + --executor-memory=5G + --conf spark.executor.memoryOverhead=3g + --conf spark.sql.shuffle.partitions=3284 --driver-memory=${sparkDriverMemory} --conf spark.extraListeners=${spark2ExtraListeners} --conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners} @@ -191,8 +208,10 @@ eu.dnetlib.dhp.resulttocommunityfromsemrel.SparkResultToCommunityThroughSemRelJob dhp-enrichment-${projectVersion}.jar - --executor-cores=${sparkExecutorCores} - --executor-memory=${sparkExecutorMemory} + --executor-cores=6 + --executor-memory=5G + --conf spark.executor.memoryOverhead=3g + --conf spark.sql.shuffle.partitions=3284 --driver-memory=${sparkDriverMemory} --conf spark.extraListeners=${spark2ExtraListeners} --conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners} @@ -220,8 +239,10 @@ eu.dnetlib.dhp.resulttocommunityfromsemrel.SparkResultToCommunityThroughSemRelJob dhp-enrichment-${projectVersion}.jar - --executor-cores=${sparkExecutorCores} - --executor-memory=${sparkExecutorMemory} + --executor-cores=6 + --executor-memory=5G + --conf spark.executor.memoryOverhead=3g + --conf spark.sql.shuffle.partitions=3284 --driver-memory=${sparkDriverMemory} --conf spark.extraListeners=${spark2ExtraListeners} --conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners} @@ -249,8 +270,10 @@ eu.dnetlib.dhp.resulttocommunityfromsemrel.SparkResultToCommunityThroughSemRelJob dhp-enrichment-${projectVersion}.jar - --executor-cores=${sparkExecutorCores} - --executor-memory=${sparkExecutorMemory} + --executor-cores=6 + --executor-memory=5G + --conf spark.executor.memoryOverhead=3g + --conf spark.sql.shuffle.partitions=3284 --driver-memory=${sparkDriverMemory} --conf spark.extraListeners=${spark2ExtraListeners} --conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners} @@ -278,8 +301,10 @@ eu.dnetlib.dhp.resulttocommunityfromsemrel.SparkResultToCommunityThroughSemRelJob dhp-enrichment-${projectVersion}.jar - --executor-cores=${sparkExecutorCores} - --executor-memory=${sparkExecutorMemory} + --executor-cores=6 + --executor-memory=5G + --conf spark.executor.memoryOverhead=3g + --conf spark.sql.shuffle.partitions=3284 --driver-memory=${sparkDriverMemory} --conf spark.extraListeners=${spark2ExtraListeners} --conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners} @@ -295,10 +320,11 @@ --outputPath${workingDir}/communitysemrel/software - + + diff --git a/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/wf/subworkflows/resulttoorganizationfrominstrepo/oozie_app/workflow.xml b/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/wf/subworkflows/resulttoorganizationfrominstrepo/oozie_app/workflow.xml index dadea2d280..e963453da9 100644 --- a/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/wf/subworkflows/resulttoorganizationfrominstrepo/oozie_app/workflow.xml +++ b/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/wf/subworkflows/resulttoorganizationfrominstrepo/oozie_app/workflow.xml @@ -21,12 +21,21 @@ - + Action failed, error message[${wf:errorMessage(wf:lastErrorNode())}] + + + + + + + + + @@ -79,7 +88,7 @@ --conf spark.dynamicAllocation.maxExecutors=${spark2MaxExecutors} --sourcePath${sourcePath}/publication - --outputPath${workingDir}/affiliationinstrepo/publication/relation + --outputPath${workingDir}/affiliationInstRepo/publication/relation --datasourceOrganizationPath${workingDir}/affiliationInstRepo/preparedInfo/datasourceOrganization --alreadyLinkedPath${workingDir}/affiliationInstRepo/preparedInfo/alreadyLinked --hive_metastore_uris${hive_metastore_uris} @@ -108,7 +117,7 @@ --conf spark.dynamicAllocation.maxExecutors=${spark2MaxExecutors} --sourcePath${sourcePath}/dataset - --outputPath${workingDir}/affiliationinstrepo/dataset/relation + --outputPath${workingDir}/affiliationInstRepo/dataset/relation --datasourceOrganizationPath${workingDir}/affiliationInstRepo/preparedInfo/datasourceOrganization --alreadyLinkedPath${workingDir}/affiliationInstRepo/preparedInfo/alreadyLinked --hive_metastore_uris${hive_metastore_uris} @@ -137,7 +146,7 @@ --conf spark.dynamicAllocation.maxExecutors=${spark2MaxExecutors} --sourcePath${sourcePath}/otherresearchproduct - --outputPath${workingDir}/affiliationinstrepo/otherresearchproduct/relation + --outputPath${workingDir}/affiliationInstRepo/otherresearchproduct/relation --datasourceOrganizationPath${workingDir}/affiliationInstRepo/preparedInfo/datasourceOrganization --alreadyLinkedPath${workingDir}/affiliationInstRepo/preparedInfo/alreadyLinked --hive_metastore_uris${hive_metastore_uris} @@ -166,7 +175,7 @@ --conf spark.dynamicAllocation.maxExecutors=${spark2MaxExecutors} --sourcePath${sourcePath}/software - --outputPath${workingDir}/affiliationinstrepo/software/relation + --outputPath${workingDir}/affiliationInstRepo/software/relation --datasourceOrganizationPath${workingDir}/affiliationInstRepo/preparedInfo/datasourceOrganization --alreadyLinkedPath${workingDir}/affiliationInstRepo/preparedInfo/alreadyLinked --hive_metastore_uris${hive_metastore_uris} @@ -197,7 +206,7 @@ --conf spark.dynamicAllocation.maxExecutors=${spark2MaxExecutors} --outputPath${sourcePath}/relation - --sourcePath${workingDir}/affiliationinstrepo/ + --sourcePath${workingDir}/affiliationInstRepo/ diff --git a/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/wf/subworkflows/resulttoorganizationfromsemrel/oozie_app/config-default.xml b/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/wf/subworkflows/resulttoorganizationfromsemrel/oozie_app/config-default.xml deleted file mode 100644 index 2744ea92ba..0000000000 --- a/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/wf/subworkflows/resulttoorganizationfromsemrel/oozie_app/config-default.xml +++ /dev/null @@ -1,58 +0,0 @@ - - - jobTracker - yarnRM - - - nameNode - hdfs://nameservice1 - - - oozie.use.system.libpath - true - - - oozie.action.sharelib.for.spark - spark2 - - - hive_metastore_uris - thrift://iis-cdh5-test-m3.ocean.icm.edu.pl:9083 - - - spark2YarnHistoryServerAddress - http://iis-cdh5-test-gw.ocean.icm.edu.pl:18089 - - - spark2EventLogDir - /user/spark/spark2ApplicationHistory - - - spark2ExtraListeners - com.cloudera.spark.lineage.NavigatorAppListener - - - spark2SqlQueryExecutionListeners - com.cloudera.spark.lineage.NavigatorQueryListener - - - sparkExecutorNumber - 4 - - - sparkDriverMemory - 15G - - - sparkExecutorMemory - 6G - - - sparkExecutorCores - 1 - - - spark2MaxExecutors - 50 - - \ No newline at end of file diff --git a/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/wf/subworkflows/resulttoorganizationfromsemrel/oozie_app/workflow.xml b/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/wf/subworkflows/resulttoorganizationfromsemrel/oozie_app/workflow.xml deleted file mode 100644 index 7918df120b..0000000000 --- a/dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/wf/subworkflows/resulttoorganizationfromsemrel/oozie_app/workflow.xml +++ /dev/null @@ -1,97 +0,0 @@ - - - - sourcePath - the source path - - - - - - ${jobTracker} - ${nameNode} - - - oozie.action.sharelib.for.spark - ${oozieActionShareLibForSpark2} - - - - - - - - Action failed, error message[${wf:errorMessage(wf:lastErrorNode())}] - - - - - yarn - cluster - PrepareResultOrganizationAssociation - eu.dnetlib.dhp.resulttoorganizationfromsemrel.PrepareInfo - dhp-enrichment-${projectVersion}.jar - - --executor-cores=${sparkExecutorCores} - --executor-memory=${sparkExecutorMemory} - --driver-memory=${sparkDriverMemory} - --conf spark.extraListeners=${spark2ExtraListeners} - --conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners} - --conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress} - --conf spark.eventLog.dir=${nameNode}${spark2EventLogDir} - - --graphPath${sourcePath} - --hive_metastore_uris${hive_metastore_uris} - --leavesPath${workingDir}/affiliationSemanticRelation/preparedInfo/leavesPath - --childParentPath${workingDir}/affiliationSemanticRelation/preparedInfo/childParentPath - --resultOrgPath${workingDir}/affiliationSemanticRelation/preparedInfo/resultOrgPath - --relationPath${workingDir}/affiliationSemanticRelation/preparedInfo/relation - - - - - - - - yarn - cluster - resultToOrganizationFromSemRel - eu.dnetlib.dhp.resulttoorganizationfromsemrel.SparkResultToOrganizationFromSemRel - dhp-enrichment-${projectVersion}.jar - - --executor-cores=${sparkExecutorCores} - --executor-memory=${sparkExecutorMemory} - --driver-memory=${sparkDriverMemory} - --conf spark.extraListeners=${spark2ExtraListeners} - --conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners} - --conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress} - --conf spark.eventLog.dir=${nameNode}${spark2EventLogDir} - --conf spark.dynamicAllocation.enabled=true - --conf spark.dynamicAllocation.maxExecutors=${spark2MaxExecutors} - --conf spark.sql.shuffle.partitions=3840 - - --relationPath${workingDir}/affiliationSemanticRelation/preparedInfo/relation - --outputPath${sourcePath} - --leavesPath${workingDir}/affiliationSemanticRelation/preparedInfo/leavesPath - --childParentPath${workingDir}/affiliationSemanticRelation/preparedInfo/childParentPath - --resultOrgPath${workingDir}/affiliationSemanticRelation/preparedInfo/resultOrgPath - --hive_metastore_uris${hive_metastore_uris} - --workingDir${workingDir}/affiliationSemanticRelation/working - --iterations${iterations} - - - - - - - - - - - - - - - - - \ No newline at end of file diff --git a/dhp-workflows/dhp-graph-mapper/src/main/resources/eu/dnetlib/dhp/oa/graph/hostedbymap/oozie_app/download.sh b/dhp-workflows/dhp-graph-mapper/src/main/resources/eu/dnetlib/dhp/oa/graph/hostedbymap/oozie_app/download.sh index 35220bd8c9..9877fe7de5 100644 --- a/dhp-workflows/dhp-graph-mapper/src/main/resources/eu/dnetlib/dhp/oa/graph/hostedbymap/oozie_app/download.sh +++ b/dhp-workflows/dhp-graph-mapper/src/main/resources/eu/dnetlib/dhp/oa/graph/hostedbymap/oozie_app/download.sh @@ -1,3 +1,3 @@ #!/bin/bash curl -LSs $1 | hdfs dfs -put - $2/$3 -curl -LSs http://api.crossref.org/works/10.1099/jgv.0.001453 > prova.txt \ No newline at end of file +#curl -LSs http://api.crossref.org/works/10.1099/jgv.0.001453 > prova.txt \ No newline at end of file