kunal642 commented on a change in pull request #4110:
URL: https://github.com/apache/carbondata/pull/4110#discussion_r607040648
##########
File path:
core/src/main/java/org/apache/carbondata/core/index/secondaryindex/CarbonCostBasedOptimizer.java
##########
@@ -15,15 +15,49 @@
* limitations under the License.
*/
-package org.apache.spark.sql.secondaryindex.optimizer;
+package org.apache.carbondata.core.index.secondaryindex;
+import java.io.IOException;
import java.util.ArrayList;
+import java.util.Arrays;
+import java.util.HashMap;
import java.util.HashSet;
import java.util.List;
import java.util.Map;
import java.util.Set;
+import org.apache.carbondata.core.constants.CarbonCommonConstants;
+import org.apache.carbondata.core.metadata.index.IndexType;
+import org.apache.carbondata.core.metadata.schema.indextable.IndexMetadata;
+import org.apache.carbondata.core.metadata.schema.table.TableInfo;
+
public class CarbonCostBasedOptimizer {
+ public static Map<String, List<String>> getSecondaryIndexes(TableInfo
tableInfo) {
Review comment:
this method should be called identifyRequiredTables(Set<String>
filterAttributes, TableInfo tableInfo)... internally you can extract the table
info string and pass to the other method to get the valid SI list
##########
File path:
core/src/main/java/org/apache/carbondata/core/util/CarbonProperties.java
##########
@@ -2187,6 +2187,24 @@ public static boolean isFilterReorderingEnabled() {
);
}
+ /**
+ * Check whether plan rewrite with secondary index is enabled or not
+ */
+ public boolean isRewritePlanWithSecondaryIndex(String dbName, String
tableName) {
+ String configuredValue = getSessionPropertyValue(
Review comment:
I feel this should be based on whether the query is from a non spark
engine
##########
File path:
core/src/main/java/org/apache/carbondata/core/constants/CarbonCommonConstants.java
##########
@@ -2316,6 +2316,12 @@ private CarbonCommonConstants() {
@CarbonProperty(dynamicConfigurable = true)
public static final String CARBON_ENABLE_INDEX_SERVER =
"carbon.enable.index.server";
+ @CarbonProperty(dynamicConfigurable = true)
+ public static final String CARBON_REWRITE_PLAN_WITH_SECONDARY_INDEX =
+ "carbon.rewrite.plan.with.secondary.index";
Review comment:
add this new property in docs
##########
File path:
integration/spark/src/main/scala/org/apache/carbondata/index/secondary/SecondaryIndex.java
##########
@@ -0,0 +1,128 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one or more
+ * contributor license agreements. See the NOTICE file distributed with
+ * this work for additional information regarding copyright ownership.
+ * The ASF licenses this file to You under the Apache License, Version 2.0
+ * (the "License"); you may not use this file except in compliance with
+ * the License. You may obtain a copy of the License at
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+package org.apache.carbondata.index.secondary;
+
+import java.util.ArrayList;
+import java.util.HashSet;
+import java.util.List;
+import java.util.Map;
+import java.util.Set;
+
+import org.apache.carbondata.common.logging.LogServiceFactory;
+import org.apache.carbondata.core.datastore.block.SegmentProperties;
+import org.apache.carbondata.core.index.IndexUtil;
+import org.apache.carbondata.core.index.dev.IndexModel;
+import org.apache.carbondata.core.index.dev.cgindex.CoarseGrainIndex;
+import org.apache.carbondata.core.indexstore.Blocklet;
+import org.apache.carbondata.core.metadata.schema.table.CarbonTable;
+import org.apache.carbondata.core.scan.expression.Expression;
+import org.apache.carbondata.core.scan.filter.executer.FilterExecutor;
+import org.apache.carbondata.core.scan.filter.resolver.FilterResolverIntf;
+import
org.apache.carbondata.index.secondary.SecondaryIndexModel.PositionReferenceInfo;
+
+import org.apache.log4j.Logger;
+
+import static
org.apache.carbondata.core.util.path.CarbonTablePath.BATCH_PREFIX;
+
+/**
+ * Secondary Index to prune at blocklet level.
+ */
+public class SecondaryIndex extends CoarseGrainIndex {
+ private static final Logger LOGGER =
+ LogServiceFactory.getLogService(SecondaryIndex.class.getName());
+ String indexName;
+ String currentSegmentId;
+ List<String> validSegmentIds;
+ PositionReferenceInfo positionReferenceInfo;
+
+ @Override
+ public void init(IndexModel indexModel) {
+ assert (indexModel instanceof SecondaryIndexModel);
+ SecondaryIndexModel model = (SecondaryIndexModel) indexModel;
+ indexName = model.indexName;
+ currentSegmentId = model.currentSegmentId;
+ validSegmentIds = model.validSegmentIds;
+ positionReferenceInfo = model.positionReferenceInfo;
+ }
+
+ private Set<String> getPositionReferences(String databaseName, String
indexName,
+ Expression expression) {
+ synchronized (positionReferenceInfo) {
Review comment:
there is no need for this as this will be a sequential call
##########
File path:
integration/spark/src/main/scala/org/apache/carbondata/indexserver/DistributedRDDUtils.scala
##########
@@ -394,4 +397,35 @@ object DistributedRDDUtils {
}
}
}
+
+ def pruneOnDriver(request: IndexInputFormat): ExtendedBlockletWrapper = {
+ val job = CarbonSparkUtil.createHadoopJob()
+ val conf = job.getConfiguration
+ val tableInfo = request.getCarbonTable.getTableInfo
+ val identifier = request.getCarbonTable.getAbsoluteTableIdentifier
+ val indexFilter = new IndexFilter(request.getCarbonTable,
+ request.getFilterResolverIntf.getFilterExpression)
+ CarbonInputFormat.setTableInfo(conf, tableInfo)
+ CarbonInputFormat.setFilterPredicates(conf, indexFilter)
+ CarbonInputFormat.setDatabaseName(conf, tableInfo.getDatabaseName)
+ CarbonInputFormat.setTableName(conf, tableInfo.getFactTable.getTableName)
+ CarbonInputFormat.setPartitionsToPrune(conf, request.getPartitions)
+ CarbonInputFormat.setTransactionalTable(conf,
tableInfo.isTransactionalTable)
+ CarbonInputFormat.setTablePath(conf,
+ identifier.appendWithLocalPrefix(identifier.getTablePath))
+ CarbonInputFormat.setQuerySegment(conf, identifier)
+ CarbonInputFormat.setColumnProjection(conf, Array("positionId"))
+ CarbonInputFormat.setReadCommittedScope(conf,
request.getReadCommittedScope)
+ CarbonInputFormat.setSegmentsToAccess(conf, request.getValidSegments)
+ CarbonInputFormat.setValidateSegmentsToAccess(conf, false)
+ CarbonInputFormat.setSecondaryIndexPruning(conf,
request.isSIPruningEnabled)
+ val blocklets = new CarbonTableInputFormat[Object].getPrunedBlocklets(job,
Review comment:
set useIndexServer as false in job and getPrunedBlocklets should check
for "useIndexServer" to decide whether to go for indexServer or not.
Otherwise if user sets indexServerEnable to true in carbon.properties for
indexserver application then this flow can go in a infinite loop.
##########
File path:
integration/spark/src/main/scala/org/apache/carbondata/index/secondary/SecondaryIndex.java
##########
@@ -0,0 +1,128 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one or more
+ * contributor license agreements. See the NOTICE file distributed with
+ * this work for additional information regarding copyright ownership.
+ * The ASF licenses this file to You under the Apache License, Version 2.0
+ * (the "License"); you may not use this file except in compliance with
+ * the License. You may obtain a copy of the License at
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+package org.apache.carbondata.index.secondary;
+
+import java.util.ArrayList;
+import java.util.HashSet;
+import java.util.List;
+import java.util.Map;
+import java.util.Set;
+
+import org.apache.carbondata.common.logging.LogServiceFactory;
+import org.apache.carbondata.core.datastore.block.SegmentProperties;
+import org.apache.carbondata.core.index.IndexUtil;
+import org.apache.carbondata.core.index.dev.IndexModel;
+import org.apache.carbondata.core.index.dev.cgindex.CoarseGrainIndex;
+import org.apache.carbondata.core.indexstore.Blocklet;
+import org.apache.carbondata.core.metadata.schema.table.CarbonTable;
+import org.apache.carbondata.core.scan.expression.Expression;
+import org.apache.carbondata.core.scan.filter.executer.FilterExecutor;
+import org.apache.carbondata.core.scan.filter.resolver.FilterResolverIntf;
+import
org.apache.carbondata.index.secondary.SecondaryIndexModel.PositionReferenceInfo;
+
+import org.apache.log4j.Logger;
+
+import static
org.apache.carbondata.core.util.path.CarbonTablePath.BATCH_PREFIX;
+
+/**
+ * Secondary Index to prune at blocklet level.
+ */
+public class SecondaryIndex extends CoarseGrainIndex {
+ private static final Logger LOGGER =
+ LogServiceFactory.getLogService(SecondaryIndex.class.getName());
+ String indexName;
+ String currentSegmentId;
+ List<String> validSegmentIds;
+ PositionReferenceInfo positionReferenceInfo;
Review comment:
Can you add details for why positionReferenceInfo is same across the
SecondaryIndex instances for each segment.
##########
File path:
integration/spark/src/main/scala/org/apache/carbondata/indexserver/IndexServer.scala
##########
@@ -265,8 +272,6 @@ object IndexServer extends ServerInterface {
server.stop()
}
})
- CarbonProperties.getInstance().addProperty(CarbonCommonConstants
Review comment:
please revert this
##########
File path:
integration/presto/src/main/prestodb/org/apache/carbondata/presto/impl/CarbonTableReader.java
##########
@@ -270,6 +270,7 @@ private CarbonTableCacheModel
getValidCacheBySchemaTableName(SchemaTableName sch
config.set("query.id", queryId);
CarbonInputFormat.setTransactionalTable(config,
carbonTable.isTransactionalTable());
CarbonInputFormat.setTableInfo(config, carbonTable.getTableInfo());
+
CarbonInputFormat.checkAndSetSecondaryIndexPruning(carbonTable.getTableInfo(),
filters, config);
Review comment:
Later we can optimize the CarbonInputSplit to return only the required
objects incase query performance is bad... Please add a TODO for same
##########
File path:
integration/spark/src/main/scala/org/apache/spark/sql/index/CarbonIndexUtil.scala
##########
@@ -396,6 +398,88 @@ object CarbonIndexUtil {
}
}
+ def updateIndexStatusInBatch(carbonTable: CarbonTable,
+ indexTables: List[CarbonTable],
+ indexType: IndexType,
+ status: IndexStatus,
+ sparkSession: SparkSession): Unit = {
+ val dbName = carbonTable.getDatabaseName
+ val tableName = carbonTable.getTableName
+ val metadataLock =
CarbonLockFactory.getCarbonLockObj(carbonTable.getAbsoluteTableIdentifier,
+ LockUsage.METADATA_LOCK)
+ try {
+ if (metadataLock.lockWithRetries()) {
+ CarbonMetadata.getInstance.removeTable(dbName, tableName)
+ val table = CarbonEnv.getCarbonTable(Some(dbName),
tableName)(sparkSession)
+ val indexTableInfos = IndexTableInfo.fromGson(table.getIndexInfo)
+ val indexMetadata = table.getIndexMetadata
+ indexTables.foreach { indexTable =>
+ IndexTableInfo.setIndexStatus(indexTableInfos,
indexTable.getTableName, status)
+ indexMetadata.updateIndexStatus(indexType.getIndexProviderName,
+ indexTable.getTableName,
+ status.name())
+ }
+ table.getTableInfo
+ .getFactTable
+ .getTableProperties
+ .put(table.getCarbonTableIdentifier.getTableId,
indexMetadata.serialize)
+ sparkSession.sql(s"""ALTER TABLE $dbName.$tableName SET
SERDEPROPERTIES ('indexInfo' = '${
+ IndexTableInfo.toGson(indexTableInfos)
+ }')""".stripMargin).collect()
+ CarbonHiveIndexMetadataUtil.refreshTable(dbName, tableName,
sparkSession)
+ CarbonMetadata.getInstance.removeTable(dbName, tableName)
+ CarbonMetadata.getInstance.loadTableMetadata(table.getTableInfo)
+ }
+ } finally {
+ metadataLock.unlock()
+ }
+ }
+
+ def updateIndexStatus(carbonTable: CarbonTable,
Review comment:
If SI is already created on a table and its in sync, then IndexStatus
would not be there as this is a new DataMap Implementation for us.
We need to find some way to update this property for old SI tables
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
[email protected]
