Search in sources :

Example 16 with PartitionInfo

use of com.netflix.metacat.common.server.connectors.model.PartitionInfo in project metacat by Netflix.

the class HiveConnectorPartitionService method getPartitionUri.

private String getPartitionUri(final PartitionHolder partition) {
    String result = null;
    if (partition.getPartition() != null) {
        final Partition hivePartition = partition.getPartition();
        result = hivePartition.getSd() != null ? hivePartition.getSd().getLocation() : null;
    } else if (partition.getPartitionInfo() != null) {
        final PartitionInfo partitionInfo = partition.getPartitionInfo();
        result = partitionInfo.getSerde() != null ? partitionInfo.getSerde().getUri() : null;
    }
    return result;
}
Also used : Partition(org.apache.hadoop.hive.metastore.api.Partition) PartitionInfo(com.netflix.metacat.common.server.connectors.model.PartitionInfo)

Example 17 with PartitionInfo

use of com.netflix.metacat.common.server.connectors.model.PartitionInfo in project metacat by Netflix.

the class HiveConnectorPartitionService method savePartitions.

/**
 * By default(checkIfExists=true and aletrIfExists=false), this method adds the provided list of partitions.
 * If a partition already exists, it is dropped first before adding it.
 * If checkIfExists=false, the method adds the partitions to the table. If a partition already exists,
 * an AlreadyExistsException error is thrown.
 * If alterIfExists=true, the method updates existing partitions and adds non-existant partitions.
 *
 * If a partition in the provided partition list has all the details, then it is used. If the details are missing,
 * then the table details are inherited. This is mostly for the storage information.
 */
@Override
public PartitionsSaveResponse savePartitions(final ConnectorRequestContext requestContext, final QualifiedName tableQName, final PartitionsSaveRequest partitionsSaveRequest) {
    final String databaseName = tableQName.getDatabaseName();
    final String tableName = tableQName.getTableName();
    final Table table;
    try {
        table = metacatHiveClient.getTableByName(databaseName, tableName);
    } catch (NoSuchObjectException exception) {
        throw new TableNotFoundException(tableQName, exception);
    } catch (TException e) {
        throw new ConnectorException(String.format("Failed getting hive table %s", tableQName), e);
    }
    // New partitions
    final List<PartitionInfo> addedPartitionInfos = Lists.newArrayList();
    final List<PartitionInfo> partitionInfos = partitionsSaveRequest.getPartitions();
    final List<String> partitionNames = partitionInfos.stream().map(part -> {
        final String partitionName = part.getName().getPartitionName();
        PartitionUtil.validatePartitionName(partitionName, getPartitionKeys(table.getPartitionKeys()));
        return partitionName;
    }).collect(Collectors.toList());
    // New partition names
    final List<String> addedPartitionNames = Lists.newArrayList();
    // Updated partition names
    final List<String> existingPartitionNames = Lists.newArrayList();
    // Existing partitions
    final List<PartitionHolder> existingPartitionHolders = Lists.newArrayList();
    // Existing partition map
    Map<String, PartitionHolder> existingPartitionMap = Collections.emptyMap();
    // 
    if (partitionsSaveRequest.getCheckIfExists() || partitionsSaveRequest.getAlterIfExists()) {
        existingPartitionMap = getPartitionsByNames(table, partitionNames);
    }
    for (PartitionInfo partitionInfo : partitionInfos) {
        final String partitionName = partitionInfo.getName().getPartitionName();
        final PartitionHolder existingPartitionHolder = existingPartitionMap.get(partitionName);
        if (existingPartitionHolder == null) {
            addedPartitionNames.add(partitionName);
            addedPartitionInfos.add(partitionInfo);
        } else {
            final String partitionUri = partitionInfo.getSerde() != null ? partitionInfo.getSerde().getUri() : null;
            final String existingPartitionUri = getPartitionUri(existingPartitionHolder);
            if (partitionUri == null || !partitionUri.equals(existingPartitionUri)) {
                existingPartitionNames.add(partitionName);
                // We need to copy the existing partition info and
                if (partitionInfo.getSerde() == null) {
                    partitionInfo.setSerde(new StorageInfo());
                }
                if (partitionInfo.getAudit() == null) {
                    partitionInfo.setAudit(new AuditInfo());
                }
                if (StringUtils.isBlank(partitionUri)) {
                    partitionInfo.getSerde().setUri(existingPartitionUri);
                }
                // unless we alterifExists
                if (partitionsSaveRequest.getAlterIfExists()) {
                    if (existingPartitionHolder.getPartition() != null) {
                        final Partition existingPartition = existingPartitionHolder.getPartition();
                        partitionInfo.getSerde().setParameters(existingPartition.getParameters());
                        partitionInfo.getAudit().setCreatedDate(HiveConnectorInfoConverter.epochSecondsToDate(existingPartition.getCreateTime()));
                        partitionInfo.getAudit().setLastModifiedDate(HiveConnectorInfoConverter.epochSecondsToDate(existingPartition.getLastAccessTime()));
                    } else {
                        final PartitionInfo existingPartitionInfo = existingPartitionHolder.getPartitionInfo();
                        if (existingPartitionInfo.getSerde() != null) {
                            partitionInfo.getSerde().setParameters(existingPartitionInfo.getSerde().getParameters());
                        }
                        if (existingPartitionInfo.getAudit() != null) {
                            partitionInfo.getAudit().setCreatedDate(existingPartitionInfo.getAudit().getCreatedDate());
                            partitionInfo.getAudit().setLastModifiedDate(existingPartitionInfo.getAudit().getLastModifiedDate());
                        }
                    }
                    existingPartitionHolder.setPartitionInfo(partitionInfo);
                    existingPartitionHolders.add(existingPartitionHolder);
                } else {
                    addedPartitionInfos.add(partitionInfo);
                }
            }
        }
    }
    final Set<String> deletePartitionNames = Sets.newHashSet();
    if (!partitionsSaveRequest.getAlterIfExists()) {
        deletePartitionNames.addAll(existingPartitionNames);
    }
    if (partitionsSaveRequest.getPartitionIdsForDeletes() != null) {
        deletePartitionNames.addAll(partitionsSaveRequest.getPartitionIdsForDeletes());
    }
    addUpdateDropPartitions(tableQName, table, partitionNames, addedPartitionInfos, existingPartitionHolders, deletePartitionNames);
    final PartitionsSaveResponse result = new PartitionsSaveResponse();
    result.setAdded(addedPartitionNames);
    result.setUpdated(existingPartitionNames);
    return result;
}
Also used : TException(org.apache.thrift.TException) StringUtils(org.apache.commons.lang.StringUtils) Getter(lombok.Getter) MetaException(org.apache.hadoop.hive.metastore.api.MetaException) SortOrder(com.netflix.metacat.common.dto.SortOrder) AuditInfo(com.netflix.metacat.common.server.connectors.model.AuditInfo) HashMap(java.util.HashMap) SerDeInfo(org.apache.hadoop.hive.metastore.api.SerDeInfo) Partition(org.apache.hadoop.hive.metastore.api.Partition) Function(java.util.function.Function) Warehouse(org.apache.hadoop.hive.metastore.Warehouse) ArrayList(java.util.ArrayList) AlreadyExistsException(org.apache.hadoop.hive.metastore.api.AlreadyExistsException) Strings(com.google.common.base.Strings) ConnectorPartitionService(com.netflix.metacat.common.server.connectors.ConnectorPartitionService) InvalidMetaException(com.netflix.metacat.common.server.connectors.exception.InvalidMetaException) Lists(com.google.common.collect.Lists) ConnectorException(com.netflix.metacat.common.server.connectors.exception.ConnectorException) PartitionInfo(com.netflix.metacat.common.server.connectors.model.PartitionInfo) Map(java.util.Map) ConnectorContext(com.netflix.metacat.common.server.connectors.ConnectorContext) StorageInfo(com.netflix.metacat.common.server.connectors.model.StorageInfo) HiveConnectorInfoConverter(com.netflix.metacat.connector.hive.converters.HiveConnectorInfoConverter) PartitionUtil(com.netflix.metacat.common.server.partition.util.PartitionUtil) StorageDescriptor(org.apache.hadoop.hive.metastore.api.StorageDescriptor) ConnectorRequestContext(com.netflix.metacat.common.server.connectors.ConnectorRequestContext) Nullable(javax.annotation.Nullable) PartitionHolder(com.netflix.metacat.connector.hive.sql.PartitionHolder) Pageable(com.netflix.metacat.common.dto.Pageable) TException(org.apache.thrift.TException) Set(java.util.Set) QualifiedName(com.netflix.metacat.common.QualifiedName) InvalidObjectException(org.apache.hadoop.hive.metastore.api.InvalidObjectException) TableNotFoundException(com.netflix.metacat.common.server.connectors.exception.TableNotFoundException) Collectors(java.util.stream.Collectors) Sets(com.google.common.collect.Sets) Table(org.apache.hadoop.hive.metastore.api.Table) PartitionsSaveResponse(com.netflix.metacat.common.server.connectors.model.PartitionsSaveResponse) FieldSchema(org.apache.hadoop.hive.metastore.api.FieldSchema) List(java.util.List) PartitionAlreadyExistsException(com.netflix.metacat.common.server.connectors.exception.PartitionAlreadyExistsException) TableInfo(com.netflix.metacat.common.server.connectors.model.TableInfo) PartitionsSaveRequest(com.netflix.metacat.common.server.connectors.model.PartitionsSaveRequest) PartitionListRequest(com.netflix.metacat.common.server.connectors.model.PartitionListRequest) ConnectorUtils(com.netflix.metacat.common.server.connectors.ConnectorUtils) PartitionNotFoundException(com.netflix.metacat.common.server.connectors.exception.PartitionNotFoundException) Collections(java.util.Collections) NoSuchObjectException(org.apache.hadoop.hive.metastore.api.NoSuchObjectException) Sort(com.netflix.metacat.common.dto.Sort) Partition(org.apache.hadoop.hive.metastore.api.Partition) PartitionHolder(com.netflix.metacat.connector.hive.sql.PartitionHolder) AuditInfo(com.netflix.metacat.common.server.connectors.model.AuditInfo) Table(org.apache.hadoop.hive.metastore.api.Table) TableNotFoundException(com.netflix.metacat.common.server.connectors.exception.TableNotFoundException) ConnectorException(com.netflix.metacat.common.server.connectors.exception.ConnectorException) StorageInfo(com.netflix.metacat.common.server.connectors.model.StorageInfo) PartitionsSaveResponse(com.netflix.metacat.common.server.connectors.model.PartitionsSaveResponse) NoSuchObjectException(org.apache.hadoop.hive.metastore.api.NoSuchObjectException) PartitionInfo(com.netflix.metacat.common.server.connectors.model.PartitionInfo)

Example 18 with PartitionInfo

use of com.netflix.metacat.common.server.connectors.model.PartitionInfo in project metacat by Netflix.

the class HiveConnectorPartitionService method addUpdateDropPartitions.

protected void addUpdateDropPartitions(final QualifiedName tableQName, final Table table, final List<String> partitionNames, final List<PartitionInfo> addedPartitionInfos, final List<PartitionHolder> existingPartitionInfos, final Set<String> deletePartitionNames) {
    final String databaseName = table.getDbName();
    final String tableName = table.getTableName();
    final TableInfo tableInfo = hiveMetacatConverters.toTableInfo(tableQName, table);
    try {
        final List<Partition> existingPartitions = existingPartitionInfos.stream().map(p -> hiveMetacatConverters.fromPartitionInfo(tableInfo, p.getPartitionInfo())).collect(Collectors.toList());
        final List<Partition> addedPartitions = addedPartitionInfos.stream().map(p -> hiveMetacatConverters.fromPartitionInfo(tableInfo, p)).collect(Collectors.toList());
        // If alterIfExists=true, then alter partitions if they already exists
        if (!existingPartitionInfos.isEmpty()) {
            copyTableSdToPartitionSd(existingPartitions, table);
            metacatHiveClient.alterPartitions(databaseName, tableName, existingPartitions);
        }
        // Copy the storage details from the table if the partition does not contain the details.
        copyTableSdToPartitionSd(addedPartitions, table);
        // Drop partitions with ids in 'deletePartitionNames' and add 'addedPartitionInfos' partitions
        metacatHiveClient.addDropPartitions(databaseName, tableName, addedPartitions, Lists.newArrayList(deletePartitionNames));
    } catch (NoSuchObjectException exception) {
        if (exception.getMessage() != null && exception.getMessage().startsWith("Partition doesn't exist")) {
            throw new PartitionNotFoundException(tableQName, "", exception);
        } else {
            throw new TableNotFoundException(tableQName, exception);
        }
    } catch (MetaException | InvalidObjectException exception) {
        throw new InvalidMetaException("One or more partitions are invalid.", exception);
    } catch (AlreadyExistsException e) {
        throw new PartitionAlreadyExistsException(tableQName, partitionNames, e);
    } catch (TException exception) {
        throw new ConnectorException(String.format("Failed savePartitions hive table %s", tableName), exception);
    }
}
Also used : StringUtils(org.apache.commons.lang.StringUtils) Getter(lombok.Getter) MetaException(org.apache.hadoop.hive.metastore.api.MetaException) SortOrder(com.netflix.metacat.common.dto.SortOrder) AuditInfo(com.netflix.metacat.common.server.connectors.model.AuditInfo) HashMap(java.util.HashMap) SerDeInfo(org.apache.hadoop.hive.metastore.api.SerDeInfo) Partition(org.apache.hadoop.hive.metastore.api.Partition) Function(java.util.function.Function) Warehouse(org.apache.hadoop.hive.metastore.Warehouse) ArrayList(java.util.ArrayList) AlreadyExistsException(org.apache.hadoop.hive.metastore.api.AlreadyExistsException) Strings(com.google.common.base.Strings) ConnectorPartitionService(com.netflix.metacat.common.server.connectors.ConnectorPartitionService) InvalidMetaException(com.netflix.metacat.common.server.connectors.exception.InvalidMetaException) Lists(com.google.common.collect.Lists) ConnectorException(com.netflix.metacat.common.server.connectors.exception.ConnectorException) PartitionInfo(com.netflix.metacat.common.server.connectors.model.PartitionInfo) Map(java.util.Map) ConnectorContext(com.netflix.metacat.common.server.connectors.ConnectorContext) StorageInfo(com.netflix.metacat.common.server.connectors.model.StorageInfo) HiveConnectorInfoConverter(com.netflix.metacat.connector.hive.converters.HiveConnectorInfoConverter) PartitionUtil(com.netflix.metacat.common.server.partition.util.PartitionUtil) StorageDescriptor(org.apache.hadoop.hive.metastore.api.StorageDescriptor) ConnectorRequestContext(com.netflix.metacat.common.server.connectors.ConnectorRequestContext) Nullable(javax.annotation.Nullable) PartitionHolder(com.netflix.metacat.connector.hive.sql.PartitionHolder) Pageable(com.netflix.metacat.common.dto.Pageable) TException(org.apache.thrift.TException) Set(java.util.Set) QualifiedName(com.netflix.metacat.common.QualifiedName) InvalidObjectException(org.apache.hadoop.hive.metastore.api.InvalidObjectException) TableNotFoundException(com.netflix.metacat.common.server.connectors.exception.TableNotFoundException) Collectors(java.util.stream.Collectors) Sets(com.google.common.collect.Sets) Table(org.apache.hadoop.hive.metastore.api.Table) PartitionsSaveResponse(com.netflix.metacat.common.server.connectors.model.PartitionsSaveResponse) FieldSchema(org.apache.hadoop.hive.metastore.api.FieldSchema) List(java.util.List) PartitionAlreadyExistsException(com.netflix.metacat.common.server.connectors.exception.PartitionAlreadyExistsException) TableInfo(com.netflix.metacat.common.server.connectors.model.TableInfo) PartitionsSaveRequest(com.netflix.metacat.common.server.connectors.model.PartitionsSaveRequest) PartitionListRequest(com.netflix.metacat.common.server.connectors.model.PartitionListRequest) ConnectorUtils(com.netflix.metacat.common.server.connectors.ConnectorUtils) PartitionNotFoundException(com.netflix.metacat.common.server.connectors.exception.PartitionNotFoundException) Collections(java.util.Collections) NoSuchObjectException(org.apache.hadoop.hive.metastore.api.NoSuchObjectException) Sort(com.netflix.metacat.common.dto.Sort) TException(org.apache.thrift.TException) Partition(org.apache.hadoop.hive.metastore.api.Partition) AlreadyExistsException(org.apache.hadoop.hive.metastore.api.AlreadyExistsException) PartitionAlreadyExistsException(com.netflix.metacat.common.server.connectors.exception.PartitionAlreadyExistsException) InvalidMetaException(com.netflix.metacat.common.server.connectors.exception.InvalidMetaException) TableNotFoundException(com.netflix.metacat.common.server.connectors.exception.TableNotFoundException) PartitionNotFoundException(com.netflix.metacat.common.server.connectors.exception.PartitionNotFoundException) ConnectorException(com.netflix.metacat.common.server.connectors.exception.ConnectorException) TableInfo(com.netflix.metacat.common.server.connectors.model.TableInfo) NoSuchObjectException(org.apache.hadoop.hive.metastore.api.NoSuchObjectException) InvalidObjectException(org.apache.hadoop.hive.metastore.api.InvalidObjectException) PartitionAlreadyExistsException(com.netflix.metacat.common.server.connectors.exception.PartitionAlreadyExistsException) MetaException(org.apache.hadoop.hive.metastore.api.MetaException) InvalidMetaException(com.netflix.metacat.common.server.connectors.exception.InvalidMetaException)

Aggregations

PartitionInfo (com.netflix.metacat.common.server.connectors.model.PartitionInfo)18 TableNotFoundException (com.netflix.metacat.common.server.connectors.exception.TableNotFoundException)13 QualifiedName (com.netflix.metacat.common.QualifiedName)10 Map (java.util.Map)10 Strings (com.google.common.base.Strings)8 Lists (com.google.common.collect.Lists)8 ConnectorPartitionService (com.netflix.metacat.common.server.connectors.ConnectorPartitionService)8 ConnectorException (com.netflix.metacat.common.server.connectors.exception.ConnectorException)8 PartitionAlreadyExistsException (com.netflix.metacat.common.server.connectors.exception.PartitionAlreadyExistsException)8 PartitionListRequest (com.netflix.metacat.common.server.connectors.model.PartitionListRequest)8 List (java.util.List)8 Collectors (java.util.stream.Collectors)8 Pageable (com.netflix.metacat.common.dto.Pageable)7 Sort (com.netflix.metacat.common.dto.Sort)7 ArrayList (java.util.ArrayList)7 Nullable (javax.annotation.Nullable)7 Partition (org.apache.hadoop.hive.metastore.api.Partition)7 ConnectorRequestContext (com.netflix.metacat.common.server.connectors.ConnectorRequestContext)6 TableInfo (com.netflix.metacat.common.server.connectors.model.TableInfo)6 HashMap (java.util.HashMap)6