Examples with StorageInfo com.netflix.metacat.common.server.connectors.model.StorageInfo used on opensource projects

Example 1 with StorageInfo

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

the class HiveConnectorFastPartitionService method getpartitions.

private List<PartitionInfo> getpartitions(@Nonnull @NonNull final String databaseName, @Nonnull @NonNull final String tableName, @Nullable final List<String> partitionIds, final String filterExpression, final Sort sort, final Pageable pageable, final boolean includePartitionDetails) {
    final FilterPartition filter = new FilterPartition();
    // batch exists
    final boolean isBatched = !Strings.isNullOrEmpty(filterExpression) && filterExpression.contains(FIELD_BATCHID);
    final boolean hasDateCreated = !Strings.isNullOrEmpty(filterExpression) && filterExpression.contains(FIELD_DATE_CREATED);
    // Handler for reading the result set
    final ResultSetHandler<List<PartitionDetail>> handler = rs -> {
        final List<PartitionDetail> result = Lists.newArrayList();
        while (rs.next()) {
            final String name = rs.getString("name");
            final String uri = rs.getString("uri");
            final long createdDate = rs.getLong(FIELD_DATE_CREATED);
            Map<String, String> values = null;
            if (hasDateCreated) {
                values = Maps.newHashMap();
                values.put(FIELD_DATE_CREATED, createdDate + "");
            }
            if (Strings.isNullOrEmpty(filterExpression) || filter.evaluatePartitionExpression(filterExpression, name, uri, isBatched, values)) {
                final Long id = rs.getLong("id");
                final Long sdId = rs.getLong("sd_id");
                final Long serdeId = rs.getLong("serde_id");
                final String inputFormat = rs.getString("input_format");
                final String outputFormat = rs.getString("output_format");
                final String serializationLib = rs.getString("slib");
                final StorageInfo storageInfo = new StorageInfo();
                storageInfo.setUri(uri);
                storageInfo.setInputFormat(inputFormat);
                storageInfo.setOutputFormat(outputFormat);
                storageInfo.setSerializationLib(serializationLib);
                final AuditInfo auditInfo = new AuditInfo();
                auditInfo.setCreatedDate(Date.from(Instant.ofEpochSecond(createdDate)));
                auditInfo.setLastModifiedDate(Date.from(Instant.ofEpochSecond(createdDate)));
                result.add(new PartitionDetail(id, sdId, serdeId, PartitionInfo.builder().name(QualifiedName.ofPartition(catalogName, databaseName, tableName, name)).auditInfo(auditInfo).serde(storageInfo).build()));
            }
        }
        return result;
    };
    final List<PartitionInfo> partitionInfos = new ArrayList<>();
    final List<PartitionDetail> partitions = getHandlerResults(databaseName, tableName, filterExpression, partitionIds, SQL_GET_PARTITIONS, handler, sort, pageable);
    if (includePartitionDetails && !partitions.isEmpty()) {
        final List<Long> partIds = Lists.newArrayListWithCapacity(partitions.size());
        final List<Long> sdIds = Lists.newArrayListWithCapacity(partitions.size());
        final List<Long> serdeIds = Lists.newArrayListWithCapacity(partitions.size());
        for (PartitionDetail partitionDetail : partitions) {
            partIds.add(partitionDetail.getId());
            sdIds.add(partitionDetail.getSdId());
            serdeIds.add(partitionDetail.getSerdeId());
        }
        final List<ListenableFuture<Void>> futures = Lists.newArrayList();
        final Map<Long, Map<String, String>> partitionParams = Maps.newHashMap();
        futures.add(threadServiceManager.getExecutor().submit(() -> populateParameters(partIds, SQL_GET_PARTITION_PARAMS, "part_id", partitionParams)));
        final Map<Long, Map<String, String>> sdParams = Maps.newHashMap();
        if (!sdIds.isEmpty()) {
            futures.add(threadServiceManager.getExecutor().submit(() -> populateParameters(sdIds, SQL_GET_SD_PARAMS, "sd_id", sdParams)));
        }
        final Map<Long, Map<String, String>> serdeParams = Maps.newHashMap();
        if (!serdeIds.isEmpty()) {
            futures.add(threadServiceManager.getExecutor().submit(() -> populateParameters(serdeIds, SQL_GET_SERDE_PARAMS, "serde_id", serdeParams)));
        }
        try {
            Futures.transform(Futures.successfulAsList(futures), Functions.constant(null)).get(1, TimeUnit.HOURS);
        } catch (Exception e) {
            Throwables.propagate(e);
        }
        for (PartitionDetail partitionDetail : partitions) {
            partitionDetail.getPartitionInfo().setMetadata(partitionParams.get(partitionDetail.getId()));
            partitionDetail.getPartitionInfo().getSerde().setParameters(sdParams.get(partitionDetail.getSdId()));
            partitionDetail.getPartitionInfo().getSerde().setSerdeInfoParameters(serdeParams.get(partitionDetail.getSerdeId()));
        }
    }
    for (PartitionDetail partitionDetail : partitions) {
        partitionInfos.add(partitionDetail.getPartitionInfo());
    }
    return partitionInfos;
}

Example 2 with StorageInfo

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

the class S3ConnectorInfoConverter method toPartitionInfo.

@Override
public PartitionInfo toPartitionInfo(final TableInfo tableInfo, final Partition partition) {
    final QualifiedName tableName = tableInfo.getName();
    final StorageInfo storageInfo = tableInfo.getSerde();
    storageInfo.setUri(partition.getUri());
    final AuditInfo auditInfo = AuditInfo.builder().createdDate(partition.getCreatedDate()).lastModifiedDate(partition.getLastUpdatedDate()).build();
    final AuditInfo tableAuditInfo = tableInfo.getAudit();
    if (tableAuditInfo != null) {
        auditInfo.setCreatedBy(tableAuditInfo.getCreatedBy());
        auditInfo.setLastModifiedBy(tableAuditInfo.getLastModifiedBy());
    }
    return PartitionInfo.builder().name(QualifiedName.ofPartition(tableName.getCatalogName(), tableName.getDatabaseName(), tableName.getTableName(), partition.getName())).serde(storageInfo).auditInfo(auditInfo).build();
}

Example 3 with StorageInfo

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

the class S3ConnectorInfoConverter method toPartitionInfo.

/**
 * Returns a partition info.
 * @param tableName table name
 * @param table s3 table
 * @param partition partition
 * @return partition info
 */
PartitionInfo toPartitionInfo(final QualifiedName tableName, final Table table, final Partition partition) {
    final StorageInfo storageInfo = toStorageInfo(table);
    storageInfo.setUri(partition.getUri());
    final AuditInfo auditInfo = AuditInfo.builder().createdDate(partition.getCreatedDate()).lastModifiedDate(partition.getLastUpdatedDate()).build();
    final AuditInfo tableAuditInfo = toAuditInfo(table);
    if (tableAuditInfo != null) {
        auditInfo.setCreatedBy(tableAuditInfo.getCreatedBy());
        auditInfo.setLastModifiedBy(tableAuditInfo.getLastModifiedBy());
    }
    return PartitionInfo.builder().name(QualifiedName.ofPartition(tableName.getCatalogName(), tableName.getDatabaseName(), tableName.getTableName(), partition.getName())).serde(storageInfo).auditInfo(auditInfo).build();
}

Example 4 with StorageInfo

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

the class DirectSqlGetPartition method getPartitions.

private List<PartitionHolder> getPartitions(final String databaseName, final String tableName, @Nullable final List<String> partitionIds, @Nullable final String filterExpression, @Nullable final Sort sort, @Nullable final Pageable pageable, final boolean includePartitionDetails, final boolean forceDisableAudit) {
    final FilterPartition filter = config.escapePartitionNameOnFilter() ? new HiveFilterPartition() : new FilterPartition();
    // batch exists
    final boolean isBatched = !Strings.isNullOrEmpty(filterExpression) && filterExpression.contains(FIELD_BATCHID);
    final boolean hasDateCreated = !Strings.isNullOrEmpty(filterExpression) && filterExpression.contains(FIELD_DATE_CREATED);
    // Handler for reading the result set
    final ResultSetExtractor<List<PartitionHolder>> handler = rs -> {
        final List<PartitionHolder> result = Lists.newArrayList();
        final QualifiedName tableQName = QualifiedName.ofTable(catalogName, databaseName, tableName);
        int noOfRows = 0;
        while (rs.next()) {
            noOfRows++;
            final String name = rs.getString("name");
            final String uri = rs.getString("uri");
            final long createdDate = rs.getLong(FIELD_DATE_CREATED);
            Map<String, String> values = null;
            if (hasDateCreated) {
                values = Maps.newHashMap();
                values.put(FIELD_DATE_CREATED, createdDate + "");
            }
            if (Strings.isNullOrEmpty(filterExpression) || filter.evaluatePartitionExpression(filterExpression, name, uri, isBatched, values)) {
                final Long id = rs.getLong("id");
                final Long sdId = rs.getLong("sd_id");
                final Long serdeId = rs.getLong("serde_id");
                final String inputFormat = rs.getString("input_format");
                final String outputFormat = rs.getString("output_format");
                final String serializationLib = rs.getString("slib");
                final StorageInfo storageInfo = new StorageInfo();
                storageInfo.setUri(uri);
                storageInfo.setInputFormat(inputFormat);
                storageInfo.setOutputFormat(outputFormat);
                storageInfo.setSerializationLib(serializationLib);
                final AuditInfo auditInfo = new AuditInfo();
                auditInfo.setCreatedDate(Date.from(Instant.ofEpochSecond(createdDate)));
                auditInfo.setLastModifiedDate(Date.from(Instant.ofEpochSecond(createdDate)));
                result.add(new PartitionHolder(id, sdId, serdeId, PartitionInfo.builder().name(QualifiedName.ofPartition(catalogName, databaseName, tableName, name)).auditInfo(auditInfo).serde(storageInfo).build()));
            }
            // Fail if the number of partitions exceeds the threshold limit.
            if (result.size() > config.getMaxPartitionsThreshold()) {
                registry.counter(registry.createId(HiveMetrics.CounterHiveGetPartitionsExceedThresholdFailure.getMetricName()).withTags(tableQName.parts())).increment();
                final String message = String.format("Number of partitions queried for table %s exceeded the threshold %d", tableQName, config.getMaxPartitionsThreshold());
                log.warn(message);
                throw new IllegalArgumentException(message);
            }
        }
        registry.gauge(registry.createId(HiveMetrics.GaugePreExpressionFilterGetPartitionsCount.getMetricName()).withTags(tableQName.parts())).set(noOfRows);
        return result;
    };
    final List<PartitionHolder> partitions = this.getHandlerResults(databaseName, tableName, filterExpression, partitionIds, SQL.SQL_GET_PARTITIONS, handler, sort, pageable, forceDisableAudit);
    if (includePartitionDetails && !partitions.isEmpty()) {
        final List<Long> partIds = Lists.newArrayListWithCapacity(partitions.size());
        final List<Long> sdIds = Lists.newArrayListWithCapacity(partitions.size());
        final List<Long> serdeIds = Lists.newArrayListWithCapacity(partitions.size());
        for (PartitionHolder partitionHolder : partitions) {
            partIds.add(partitionHolder.getId());
            sdIds.add(partitionHolder.getSdId());
            serdeIds.add(partitionHolder.getSerdeId());
        }
        final List<ListenableFuture<Void>> futures = Lists.newArrayList();
        final Map<Long, Map<String, String>> partitionParams = Maps.newHashMap();
        futures.add(threadServiceManager.getExecutor().submit(() -> populateParameters(partIds, SQL.SQL_GET_PARTITION_PARAMS, "part_id", partitionParams)));
        final Map<Long, Map<String, String>> sdParams = Maps.newHashMap();
        if (!sdIds.isEmpty()) {
            futures.add(threadServiceManager.getExecutor().submit(() -> populateParameters(sdIds, SQL.SQL_GET_SD_PARAMS, "sd_id", sdParams)));
        }
        final Map<Long, Map<String, String>> serdeParams = Maps.newHashMap();
        if (!serdeIds.isEmpty()) {
            futures.add(threadServiceManager.getExecutor().submit(() -> populateParameters(serdeIds, SQL.SQL_GET_SERDE_PARAMS, "serde_id", serdeParams)));
        }
        ListenableFuture<List<Void>> future = null;
        try {
            future = Futures.allAsList(futures);
            final int getPartitionsDetailsTimeout = Integer.parseInt(configuration.getOrDefault(HiveConfigConstants.GET_PARTITION_DETAILS_TIMEOUT, "120"));
            future.get(getPartitionsDetailsTimeout, TimeUnit.SECONDS);
        } catch (InterruptedException | ExecutionException | TimeoutException e) {
            try {
                if (future != null) {
                    future.cancel(true);
                }
            } catch (Exception ignored) {
                log.warn("Failed cancelling the task that gets the partition details.");
            }
            Throwables.propagate(e);
        }
        for (PartitionHolder partitionHolder : partitions) {
            partitionHolder.getPartitionInfo().setMetadata(partitionParams.get(partitionHolder.getId()));
            partitionHolder.getPartitionInfo().getSerde().setParameters(sdParams.get(partitionHolder.getSdId()));
            partitionHolder.getPartitionInfo().getSerde().setSerdeInfoParameters(serdeParams.get(partitionHolder.getSerdeId()));
        }
    }
    return partitions;
}

Example 5 with StorageInfo

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

the class DirectSqlSavePartition method _insert.

@SuppressWarnings("checkstyle:methodname")
private void _insert(final QualifiedName tableQName, final Table table, final TableSequenceIds tableSequenceIds, final PartitionSequenceIds partitionSequenceIds, final List<PartitionInfo> partitions, final long currentTimeInEpoch, final int index) {
    final List<Object[]> serdesValues = Lists.newArrayList();
    final List<Object[]> serdeParamsValues = Lists.newArrayList();
    final List<Object[]> sdsValues = Lists.newArrayList();
    final List<Object[]> partitionsValues = Lists.newArrayList();
    final List<Object[]> partitionParamsValues = Lists.newArrayList();
    final List<Object[]> partitionKeyValsValues = Lists.newArrayList();
    final List<String> partitionNames = Lists.newArrayList();
    int currentIndex = index;
    for (PartitionInfo partition : partitions) {
        final StorageInfo storageInfo = partition.getSerde();
        final long partId = partitionSequenceIds.getPartId() + currentIndex;
        final long sdsId = partitionSequenceIds.getSdsId() + currentIndex;
        final long serdeId = partitionSequenceIds.getSerdeId() + currentIndex;
        final String partitionName = partition.getName().getPartitionName();
        final List<String> partValues = PartitionUtil.getPartValuesFromPartName(tableQName, table, partitionName);
        final String escapedPartName = PartitionUtil.makePartName(table.getPartitionKeys(), partValues);
        partitionsValues.add(new Object[] { 0, tableSequenceIds.getTableId(), currentTimeInEpoch, sdsId, escapedPartName, partId });
        for (int i = 0; i < partValues.size(); i++) {
            partitionKeyValsValues.add(new Object[] { partId, partValues.get(i), i });
        }
        // Partition parameters
        final Map<String, String> parameters = partition.getMetadata();
        if (parameters != null) {
            parameters.forEach((key, value) -> partitionParamsValues.add(new Object[] { value, partId, key }));
        }
        partitionParamsValues.add(new Object[] { currentTimeInEpoch, partId, PARAM_LAST_DDL_TIME });
        if (storageInfo != null) {
            serdesValues.add(new Object[] { null, storageInfo.getSerializationLib(), serdeId });
            final Map<String, String> serdeInfoParameters = storageInfo.getSerdeInfoParameters();
            if (serdeInfoParameters != null) {
                serdeInfoParameters.forEach((key, value) -> serdeParamsValues.add(new Object[] { value, serdeId, key }));
            }
            sdsValues.add(new Object[] { storageInfo.getOutputFormat(), false, tableSequenceIds.getCdId(), false, serdeId, storageInfo.getUri(), storageInfo.getInputFormat(), 0, sdsId });
        }
        partitionNames.add(partitionName);
        currentIndex++;
    }
    try {
        jdbcTemplate.batchUpdate(SQL.SERDES_INSERT, serdesValues, new int[] { Types.VARCHAR, Types.VARCHAR, Types.BIGINT });
        jdbcTemplate.batchUpdate(SQL.SERDE_PARAMS_INSERT, serdeParamsValues, new int[] { Types.VARCHAR, Types.BIGINT, Types.VARCHAR });
        jdbcTemplate.batchUpdate(SQL.SDS_INSERT, sdsValues, new int[] { Types.VARCHAR, Types.BOOLEAN, Types.BIGINT, Types.BOOLEAN, Types.BIGINT, Types.VARCHAR, Types.VARCHAR, Types.INTEGER, Types.BIGINT });
        jdbcTemplate.batchUpdate(SQL.PARTITIONS_INSERT, partitionsValues, new int[] { Types.INTEGER, Types.BIGINT, Types.INTEGER, Types.BIGINT, Types.VARCHAR, Types.BIGINT });
        jdbcTemplate.batchUpdate(SQL.PARTITION_PARAMS_INSERT, partitionParamsValues, new int[] { Types.VARCHAR, Types.BIGINT, Types.VARCHAR });
        jdbcTemplate.batchUpdate(SQL.PARTITION_KEY_VALS_INSERT, partitionKeyValsValues, new int[] { Types.BIGINT, Types.VARCHAR, Types.INTEGER });
    } catch (DuplicateKeyException e) {
        throw new PartitionAlreadyExistsException(tableQName, partitionNames, e);
    } catch (Exception e) {
        throw new ConnectorException(String.format("Failed inserting partitions %s for table %s", partitionNames, tableQName), e);
    }
}