From 216975c4ecad5cbe48533b75af9ba714ece4dab0 Mon Sep 17 00:00:00 2001 From: Claudio Atzori Date: Thu, 25 Jun 2020 12:55:52 +0200 Subject: [PATCH] restored complete provision workflow --- .../dhp/oa/provision/oozie_app/workflow.xml | 480 +++++++++++++++++- 1 file changed, 477 insertions(+), 3 deletions(-) diff --git a/dhp-workflows/dhp-graph-provision/src/main/resources/eu/dnetlib/dhp/oa/provision/oozie_app/workflow.xml b/dhp-workflows/dhp-graph-provision/src/main/resources/eu/dnetlib/dhp/oa/provision/oozie_app/workflow.xml index e98cbbc73..0d5121cf1 100644 --- a/dhp-workflows/dhp-graph-provision/src/main/resources/eu/dnetlib/dhp/oa/provision/oozie_app/workflow.xml +++ b/dhp-workflows/dhp-graph-provision/src/main/resources/eu/dnetlib/dhp/oa/provision/oozie_app/workflow.xml @@ -97,7 +97,18 @@ - + + + + + ${wf:conf('resumeFrom') eq 'prepare_relations'} + ${wf:conf('resumeFrom') eq 'fork_join_related_entities'} + ${wf:conf('resumeFrom') eq 'fork_join_all_entities'} + ${wf:conf('resumeFrom') eq 'convert_to_xml'} + ${wf:conf('resumeFrom') eq 'to_solr_index'} + + + Action failed, error message[${wf:errorMessage(wf:lastErrorNode())}] @@ -124,12 +135,475 @@ --outputPath${workingDir}/relation --relPartitions5000 + + + + + + + + + + + + + + + + + yarn + cluster + Join[relation.target = publication.id] + eu.dnetlib.dhp.oa.provision.CreateRelatedEntitiesJob_phase1 + dhp-graph-provision-${projectVersion}.jar + + --executor-cores=${sparkExecutorCoresForJoining} + --executor-memory=${sparkExecutorMemoryForJoining} + --driver-memory=${sparkDriverMemoryForJoining} + --conf spark.extraListeners=${spark2ExtraListeners} + --conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners} + --conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress} + --conf spark.eventLog.dir=${nameNode}${spark2EventLogDir} + --conf spark.sql.shuffle.partitions=7680 + --conf spark.network.timeout=${sparkNetworkTimeout} + + --inputRelationsPath${workingDir}/relation + --inputEntityPath${inputGraphRootPath}/publication + --graphTableClassNameeu.dnetlib.dhp.schema.oaf.Publication + --outputPath${workingDir}/join_partial/publication + + + + + + + + yarn + cluster + Join[relation.target = dataset.id] + eu.dnetlib.dhp.oa.provision.CreateRelatedEntitiesJob_phase1 + dhp-graph-provision-${projectVersion}.jar + + --executor-cores=${sparkExecutorCoresForJoining} + --executor-memory=${sparkExecutorMemoryForJoining} + --driver-memory=${sparkDriverMemoryForJoining} + --conf spark.extraListeners=${spark2ExtraListeners} + --conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners} + --conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress} + --conf spark.eventLog.dir=${nameNode}${spark2EventLogDir} + --conf spark.sql.shuffle.partitions=3840 + --conf spark.network.timeout=${sparkNetworkTimeout} + + --inputRelationsPath${workingDir}/relation + --inputEntityPath${inputGraphRootPath}/dataset + --graphTableClassNameeu.dnetlib.dhp.schema.oaf.Dataset + --outputPath${workingDir}/join_partial/dataset + + + + + + + + yarn + cluster + Join[relation.target = otherresearchproduct.id] + eu.dnetlib.dhp.oa.provision.CreateRelatedEntitiesJob_phase1 + dhp-graph-provision-${projectVersion}.jar + + --executor-cores=${sparkExecutorCoresForJoining} + --executor-memory=${sparkExecutorMemoryForJoining} + --driver-memory=${sparkDriverMemoryForJoining} + --conf spark.extraListeners=${spark2ExtraListeners} + --conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners} + --conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress} + --conf spark.eventLog.dir=${nameNode}${spark2EventLogDir} + --conf spark.sql.shuffle.partitions=3840 + --conf spark.network.timeout=${sparkNetworkTimeout} + + --inputRelationsPath${workingDir}/relation + --inputEntityPath${inputGraphRootPath}/otherresearchproduct + --graphTableClassNameeu.dnetlib.dhp.schema.oaf.OtherResearchProduct + --outputPath${workingDir}/join_partial/otherresearchproduct + + + + + + + + yarn + cluster + Join[relation.target = software.id] + eu.dnetlib.dhp.oa.provision.CreateRelatedEntitiesJob_phase1 + dhp-graph-provision-${projectVersion}.jar + + --executor-cores=${sparkExecutorCoresForJoining} + --executor-memory=${sparkExecutorMemoryForJoining} + --driver-memory=${sparkDriverMemoryForJoining} + --conf spark.extraListeners=${spark2ExtraListeners} + --conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners} + --conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress} + --conf spark.eventLog.dir=${nameNode}${spark2EventLogDir} + --conf spark.sql.shuffle.partitions=3840 + --conf spark.network.timeout=${sparkNetworkTimeout} + + --inputRelationsPath${workingDir}/relation + --inputEntityPath${inputGraphRootPath}/software + --graphTableClassNameeu.dnetlib.dhp.schema.oaf.Software + --outputPath${workingDir}/join_partial/software + + + + + + + + yarn + cluster + Join[relation.target = datasource.id] + eu.dnetlib.dhp.oa.provision.CreateRelatedEntitiesJob_phase1 + dhp-graph-provision-${projectVersion}.jar + + --executor-cores=${sparkExecutorCoresForJoining} + --executor-memory=${sparkExecutorMemoryForJoining} + --driver-memory=${sparkDriverMemoryForJoining} + --conf spark.extraListeners=${spark2ExtraListeners} + --conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners} + --conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress} + --conf spark.eventLog.dir=${nameNode}${spark2EventLogDir} + --conf spark.sql.shuffle.partitions=3840 + --conf spark.network.timeout=${sparkNetworkTimeout} + + --inputRelationsPath${workingDir}/relation + --inputEntityPath${inputGraphRootPath}/datasource + --graphTableClassNameeu.dnetlib.dhp.schema.oaf.Datasource + --outputPath${workingDir}/join_partial/datasource + + + + + + + + yarn + cluster + Join[relation.target = organization.id] + eu.dnetlib.dhp.oa.provision.CreateRelatedEntitiesJob_phase1 + dhp-graph-provision-${projectVersion}.jar + + --executor-cores=${sparkExecutorCoresForJoining} + --executor-memory=${sparkExecutorMemoryForJoining} + --driver-memory=${sparkDriverMemoryForJoining} + --conf spark.extraListeners=${spark2ExtraListeners} + --conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners} + --conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress} + --conf spark.eventLog.dir=${nameNode}${spark2EventLogDir} + --conf spark.sql.shuffle.partitions=3840 + --conf spark.network.timeout=${sparkNetworkTimeout} + + --inputRelationsPath${workingDir}/relation + --inputEntityPath${inputGraphRootPath}/organization + --graphTableClassNameeu.dnetlib.dhp.schema.oaf.Organization + --outputPath${workingDir}/join_partial/organization + + + + + + + + yarn + cluster + Join[relation.target = project.id] + eu.dnetlib.dhp.oa.provision.CreateRelatedEntitiesJob_phase1 + dhp-graph-provision-${projectVersion}.jar + + --executor-cores=${sparkExecutorCoresForJoining} + --executor-memory=${sparkExecutorMemoryForJoining} + --driver-memory=${sparkDriverMemoryForJoining} + --conf spark.extraListeners=${spark2ExtraListeners} + --conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners} + --conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress} + --conf spark.eventLog.dir=${nameNode}${spark2EventLogDir} + --conf spark.sql.shuffle.partitions=3840 + --conf spark.network.timeout=${sparkNetworkTimeout} + + --inputRelationsPath${workingDir}/relation + --inputEntityPath${inputGraphRootPath}/project + --graphTableClassNameeu.dnetlib.dhp.schema.oaf.Project + --outputPath${workingDir}/join_partial/project + + + + + + + + + + + + + + + + + + + + yarn + cluster + Join[publication.id = relatedEntity.source] + eu.dnetlib.dhp.oa.provision.CreateRelatedEntitiesJob_phase2 + dhp-graph-provision-${projectVersion}.jar + + --executor-cores=${sparkExecutorCoresForJoining} + --executor-memory=${sparkExecutorMemoryForJoining} + --driver-memory=${sparkDriverMemoryForJoining} + --conf spark.extraListeners=${spark2ExtraListeners} + --conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners} + --conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress} + --conf spark.eventLog.dir=${nameNode}${spark2EventLogDir} + --conf spark.sql.shuffle.partitions=15360 + --conf spark.network.timeout=${sparkNetworkTimeout} + + --inputEntityPath${inputGraphRootPath}/publication + --graphTableClassNameeu.dnetlib.dhp.schema.oaf.Publication + --inputRelatedEntitiesPath${workingDir}/join_partial + --outputPath${workingDir}/join_entities/publication + --numPartitions30000 + + + + + + + + yarn + cluster + Join[dataset.id = relatedEntity.source] + eu.dnetlib.dhp.oa.provision.CreateRelatedEntitiesJob_phase2 + dhp-graph-provision-${projectVersion}.jar + + --executor-cores=${sparkExecutorCoresForJoining} + --executor-memory=${sparkExecutorMemoryForJoining} + --driver-memory=${sparkDriverMemoryForJoining} + --conf spark.extraListeners=${spark2ExtraListeners} + --conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners} + --conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress} + --conf spark.eventLog.dir=${nameNode}${spark2EventLogDir} + --conf spark.sql.shuffle.partitions=7680 + --conf spark.network.timeout=${sparkNetworkTimeout} + + --inputEntityPath${inputGraphRootPath}/dataset + --graphTableClassNameeu.dnetlib.dhp.schema.oaf.Dataset + --inputRelatedEntitiesPath${workingDir}/join_partial + --outputPath${workingDir}/join_entities/dataset + --numPartitions20000 + + + + + + + + yarn + cluster + Join[otherresearchproduct.id = relatedEntity.source] + eu.dnetlib.dhp.oa.provision.CreateRelatedEntitiesJob_phase2 + dhp-graph-provision-${projectVersion}.jar + + --executor-cores=${sparkExecutorCoresForJoining} + --executor-memory=${sparkExecutorMemoryForJoining} + --driver-memory=${sparkDriverMemoryForJoining} + --conf spark.extraListeners=${spark2ExtraListeners} + --conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners} + --conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress} + --conf spark.eventLog.dir=${nameNode}${spark2EventLogDir} + --conf spark.sql.shuffle.partitions=7680 + --conf spark.network.timeout=${sparkNetworkTimeout} + + --inputEntityPath${inputGraphRootPath}/otherresearchproduct + --graphTableClassNameeu.dnetlib.dhp.schema.oaf.OtherResearchProduct + --inputRelatedEntitiesPath${workingDir}/join_partial + --outputPath${workingDir}/join_entities/otherresearchproduct + --numPartitions10000 + + + + + + + + yarn + cluster + Join[software.id = relatedEntity.source] + eu.dnetlib.dhp.oa.provision.CreateRelatedEntitiesJob_phase2 + dhp-graph-provision-${projectVersion}.jar + + --executor-cores=${sparkExecutorCoresForJoining} + --executor-memory=${sparkExecutorMemoryForJoining} + --driver-memory=${sparkDriverMemoryForJoining} + --conf spark.extraListeners=${spark2ExtraListeners} + --conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners} + --conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress} + --conf spark.eventLog.dir=${nameNode}${spark2EventLogDir} + --conf spark.sql.shuffle.partitions=3840 + --conf spark.network.timeout=${sparkNetworkTimeout} + + --inputEntityPath${inputGraphRootPath}/software + --graphTableClassNameeu.dnetlib.dhp.schema.oaf.Software + --inputRelatedEntitiesPath${workingDir}/join_partial + --outputPath${workingDir}/join_entities/software + --numPartitions10000 + + + + + + + + yarn + cluster + Join[datasource.id = relatedEntity.source] + eu.dnetlib.dhp.oa.provision.CreateRelatedEntitiesJob_phase2 + dhp-graph-provision-${projectVersion}.jar + + --executor-cores=${sparkExecutorCoresForJoining} + --executor-memory=${sparkExecutorMemoryForJoining} + --driver-memory=${sparkDriverMemoryForJoining} + --conf spark.extraListeners=${spark2ExtraListeners} + --conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners} + --conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress} + --conf spark.eventLog.dir=${nameNode}${spark2EventLogDir} + --conf spark.sql.shuffle.partitions=7680 + --conf spark.network.timeout=${sparkNetworkTimeout} + + --inputEntityPath${inputGraphRootPath}/datasource + --graphTableClassNameeu.dnetlib.dhp.schema.oaf.Datasource + --inputRelatedEntitiesPath${workingDir}/join_partial + --outputPath${workingDir}/join_entities/datasource + --numPartitions1000 + + + + + + + + yarn + cluster + Join[organization.id = relatedEntity.source] + eu.dnetlib.dhp.oa.provision.CreateRelatedEntitiesJob_phase2 + dhp-graph-provision-${projectVersion}.jar + + --executor-cores=${sparkExecutorCoresForJoining} + --executor-memory=${sparkExecutorMemoryForJoining} + --driver-memory=${sparkDriverMemoryForJoining} + --conf spark.extraListeners=${spark2ExtraListeners} + --conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners} + --conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress} + --conf spark.eventLog.dir=${nameNode}${spark2EventLogDir} + --conf spark.sql.shuffle.partitions=7680 + --conf spark.network.timeout=${sparkNetworkTimeout} + + --inputEntityPath${inputGraphRootPath}/organization + --graphTableClassNameeu.dnetlib.dhp.schema.oaf.Organization + --inputRelatedEntitiesPath${workingDir}/join_partial + --outputPath${workingDir}/join_entities/organization + --numPartitions20000 + + + + + + + + yarn + cluster + Join[project.id = relatedEntity.source] + eu.dnetlib.dhp.oa.provision.CreateRelatedEntitiesJob_phase2 + dhp-graph-provision-${projectVersion}.jar + + --executor-cores=${sparkExecutorCoresForJoining} + --executor-memory=${sparkExecutorMemoryForJoining} + --driver-memory=${sparkDriverMemoryForJoining} + --conf spark.extraListeners=${spark2ExtraListeners} + --conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners} + --conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress} + --conf spark.eventLog.dir=${nameNode}${spark2EventLogDir} + --conf spark.sql.shuffle.partitions=3840 + --conf spark.network.timeout=${sparkNetworkTimeout} + + --inputEntityPath${inputGraphRootPath}/project + --graphTableClassNameeu.dnetlib.dhp.schema.oaf.Project + --inputRelatedEntitiesPath${workingDir}/join_partial + --outputPath${workingDir}/join_entities/project + --numPartitions10000 + + + + + + + + + + yarn + cluster + convert_to_xml + eu.dnetlib.dhp.oa.provision.XmlConverterJob + dhp-graph-provision-${projectVersion}.jar + + --executor-cores=${sparkExecutorCoresForJoining} + --executor-memory=${sparkExecutorMemoryForJoining} + --driver-memory=${sparkDriverMemoryForJoining} + --conf spark.extraListeners=${spark2ExtraListeners} + --conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners} + --conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress} + --conf spark.eventLog.dir=${nameNode}${spark2EventLogDir} + --conf spark.sql.shuffle.partitions=3840 + --conf spark.network.timeout=${sparkNetworkTimeout} + + --inputPath${workingDir}/join_entities + --outputPath${workingDir}/xml + --isLookupUrl${isLookupUrl} + --otherDsTypeId${otherDsTypeId} + + + + + + + + yarn + cluster + to_solr_index + eu.dnetlib.dhp.oa.provision.XmlIndexingJob + dhp-graph-provision-${projectVersion}.jar + + --executor-memory=${sparkExecutorMemoryForIndexing} + --driver-memory=${sparkDriverMemoryForIndexing} + --conf spark.dynamicAllocation.enabled=true + --conf spark.dynamicAllocation.maxExecutors=${sparkExecutorCoresForIndexing} + --conf spark.extraListeners=${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 + + --inputPath${workingDir}/xml + --isLookupUrl${isLookupUrl} + --format${format} + --batchSize${batchSize} + - - \ No newline at end of file