lanyuanxiaoyao/hudi
1
0
Fork 0
Code Activity

[HUDI-2161] Adding support to disable meta columns with bulk insert operation (#3247)

Browse Source
This commit is contained in:
Sivabalan Narayanan
2021-07-19 20:43:48 -04:00
committed by GitHub
parent 2099bf41db
commit d5026e9a24
53 changed files with 1063 additions and 269 deletions
Download Patch File Download Diff File Expand all files Collapse all files

5
hudi-spark-datasource/hudi-spark3/src/main/java/org/apache/hudi/spark3/internal/DefaultSource.java
View File

@@ -19,6 +19,7 @@
package org.apache.hudi.spark3.internal;
import org.apache.hudi.DataSourceUtils;
import org.apache.hudi.common.table.HoodieTableConfig;
import org.apache.hudi.config.HoodieInternalConfig;
import org.apache.hudi.config.HoodieWriteConfig;
import org.apache.hudi.internal.BaseDefaultSource;
@@ -48,11 +49,13 @@ public class DefaultSource extends BaseDefaultSource implements TableProvider {
String instantTime = properties.get(DataSourceInternalWriterHelper.INSTANT_TIME_OPT_KEY);
String path = properties.get("path");
String tblName = properties.get(HoodieWriteConfig.TABLE_NAME.key());
boolean populateMetaFields = Boolean.parseBoolean(properties.getOrDefault(HoodieTableConfig.HOODIE_POPULATE_META_FIELDS.key(),
HoodieTableConfig.HOODIE_POPULATE_META_FIELDS.defaultValue()));
boolean arePartitionRecordsSorted = Boolean.parseBoolean(properties.getOrDefault(HoodieInternalConfig.BULKINSERT_ARE_PARTITIONER_RECORDS_SORTED,
Boolean.toString(HoodieInternalConfig.DEFAULT_BULKINSERT_ARE_PARTITIONER_RECORDS_SORTED)));
// 1st arg to createHooodieConfig is not really reuqired to be set. but passing it anyways.
HoodieWriteConfig config = DataSourceUtils.createHoodieConfig(properties.get(HoodieWriteConfig.AVRO_SCHEMA.key()), path, tblName, properties);
return new HoodieDataSourceInternalTable(instantTime, config, schema, getSparkSession(),
getConfiguration(), properties, arePartitionRecordsSorted);
getConfiguration(), properties, populateMetaFields, arePartitionRecordsSorted);
}
}

7
hudi-spark-datasource/hudi-spark3/src/main/java/org/apache/hudi/spark3/internal/HoodieBulkInsertDataInternalWriter.java
View File

@@ -37,9 +37,10 @@ public class HoodieBulkInsertDataInternalWriter implements DataWriter<InternalRo
private final BulkInsertDataInternalWriterHelper bulkInsertWriterHelper;
public HoodieBulkInsertDataInternalWriter(HoodieTable hoodieTable, HoodieWriteConfig writeConfig,
String instantTime, int taskPartitionId, long taskId, StructType structType, boolean arePartitionRecordsSorted) {
String instantTime, int taskPartitionId, long taskId, StructType structType, boolean populateMetaFields,
boolean arePartitionRecordsSorted) {
this.bulkInsertWriterHelper = new BulkInsertDataInternalWriterHelper(hoodieTable,
writeConfig, instantTime, taskPartitionId, taskId, 0, structType, arePartitionRecordsSorted);
writeConfig, instantTime, taskPartitionId, taskId, 0, structType, populateMetaFields, arePartitionRecordsSorted);
}
@Override
@@ -61,4 +62,4 @@ public class HoodieBulkInsertDataInternalWriter implements DataWriter<InternalRo
public void close() throws IOException {
bulkInsertWriterHelper.close();
}
}
}

7
hudi-spark-datasource/hudi-spark3/src/main/java/org/apache/hudi/spark3/internal/HoodieBulkInsertDataInternalWriterFactory.java
View File

@@ -35,20 +35,23 @@ public class HoodieBulkInsertDataInternalWriterFactory implements DataWriterFact
private final HoodieTable hoodieTable;
private final HoodieWriteConfig writeConfig;
private final StructType structType;
private final boolean populateMetaFields;
private final boolean arePartitionRecordsSorted;
public HoodieBulkInsertDataInternalWriterFactory(HoodieTable hoodieTable, HoodieWriteConfig writeConfig,
String instantTime, StructType structType, boolean arePartitionRecordsSorted) {
String instantTime, StructType structType, boolean populateMetaFields,
boolean arePartitionRecordsSorted) {
this.hoodieTable = hoodieTable;
this.writeConfig = writeConfig;
this.instantTime = instantTime;
this.structType = structType;
this.populateMetaFields = populateMetaFields;
this.arePartitionRecordsSorted = arePartitionRecordsSorted;
}
@Override
public DataWriter<InternalRow> createWriter(int partitionId, long taskId) {
return new HoodieBulkInsertDataInternalWriter(hoodieTable, writeConfig, instantTime, partitionId, taskId,
structType, arePartitionRecordsSorted);
structType, populateMetaFields, arePartitionRecordsSorted);
}
}

10
hudi-spark-datasource/hudi-spark3/src/main/java/org/apache/hudi/spark3/internal/HoodieDataSourceInternalBatchWrite.java
View File

@@ -27,10 +27,10 @@ import org.apache.hudi.internal.DataSourceInternalWriterHelper;
import org.apache.hadoop.conf.Configuration;
import org.apache.spark.sql.SparkSession;
import org.apache.spark.sql.connector.write.DataWriterFactory;
import org.apache.spark.sql.connector.write.WriterCommitMessage;
import org.apache.spark.sql.connector.write.BatchWrite;
import org.apache.spark.sql.connector.write.DataWriterFactory;
import org.apache.spark.sql.connector.write.PhysicalWriteInfo;
import org.apache.spark.sql.connector.write.WriterCommitMessage;
import org.apache.spark.sql.types.StructType;
import java.util.Arrays;
@@ -49,14 +49,16 @@ public class HoodieDataSourceInternalBatchWrite implements BatchWrite {
private final HoodieWriteConfig writeConfig;
private final StructType structType;
private final boolean arePartitionRecordsSorted;
private final boolean populateMetaFields;
private final DataSourceInternalWriterHelper dataSourceInternalWriterHelper;
private Map<String, String> extraMetadata = new HashMap<>();
public HoodieDataSourceInternalBatchWrite(String instantTime, HoodieWriteConfig writeConfig, StructType structType,
SparkSession jss, Configuration hadoopConfiguration, Map<String, String> properties, boolean arePartitionRecordsSorted) {
SparkSession jss, Configuration hadoopConfiguration, Map<String, String> properties, boolean populateMetaFields, boolean arePartitionRecordsSorted) {
this.instantTime = instantTime;
this.writeConfig = writeConfig;
this.structType = structType;
this.populateMetaFields = populateMetaFields;
this.arePartitionRecordsSorted = arePartitionRecordsSorted;
this.extraMetadata = DataSourceUtils.getExtraMetadata(properties);
this.dataSourceInternalWriterHelper = new DataSourceInternalWriterHelper(instantTime, writeConfig, structType,
@@ -68,7 +70,7 @@ public class HoodieDataSourceInternalBatchWrite implements BatchWrite {
dataSourceInternalWriterHelper.createInflightCommit();
if (WriteOperationType.BULK_INSERT == dataSourceInternalWriterHelper.getWriteOperationType()) {
return new HoodieBulkInsertDataInternalWriterFactory(dataSourceInternalWriterHelper.getHoodieTable(),
writeConfig, instantTime, structType, arePartitionRecordsSorted);
writeConfig, instantTime, structType, populateMetaFields, arePartitionRecordsSorted);
} else {
throw new IllegalArgumentException("Write Operation Type + " + dataSourceInternalWriterHelper.getWriteOperationType() + " not supported ");
}

11
hudi-spark-datasource/hudi-spark3/src/main/java/org/apache/hudi/spark3/internal/HoodieDataSourceInternalBatchWriteBuilder.java
View File

@@ -39,23 +39,26 @@ public class HoodieDataSourceInternalBatchWriteBuilder implements WriteBuilder {
private final StructType structType;
private final SparkSession jss;
private final Configuration hadoopConfiguration;
private final boolean arePartitionRecordsSorted;
private final Map<String, String> properties;
private final boolean populateMetaFields;
private final boolean arePartitionRecordsSorted;
public HoodieDataSourceInternalBatchWriteBuilder(String instantTime, HoodieWriteConfig writeConfig, StructType structType,
SparkSession jss, Configuration hadoopConfiguration, Map<String, String> properties, boolean arePartitionRecordsSorted) {
SparkSession jss, Configuration hadoopConfiguration, Map<String, String> properties, boolean populateMetaFields,
boolean arePartitionRecordsSorted) {
this.instantTime = instantTime;
this.writeConfig = writeConfig;
this.structType = structType;
this.jss = jss;
this.hadoopConfiguration = hadoopConfiguration;
this.arePartitionRecordsSorted = arePartitionRecordsSorted;
this.properties = properties;
this.populateMetaFields = populateMetaFields;
this.arePartitionRecordsSorted = arePartitionRecordsSorted;
}
@Override
public BatchWrite buildForBatch() {
return new HoodieDataSourceInternalBatchWrite(instantTime, writeConfig, structType, jss,
hadoopConfiguration, properties, arePartitionRecordsSorted);
hadoopConfiguration, properties, populateMetaFields, arePartitionRecordsSorted);
}
}

13
hudi-spark-datasource/hudi-spark3/src/main/java/org/apache/hudi/spark3/internal/HoodieDataSourceInternalTable.java
View File

@@ -18,9 +18,9 @@
package org.apache.hudi.spark3.internal;
import org.apache.hadoop.conf.Configuration;
import org.apache.hudi.config.HoodieWriteConfig;
import org.apache.hadoop.conf.Configuration;
import org.apache.spark.sql.SparkSession;
import org.apache.spark.sql.connector.catalog.SupportsWrite;
import org.apache.spark.sql.connector.catalog.TableCapability;
@@ -44,17 +44,19 @@ class HoodieDataSourceInternalTable implements SupportsWrite {
private final Configuration hadoopConfiguration;
private final boolean arePartitionRecordsSorted;
private final Map<String, String> properties;
private final boolean populateMetaFields;
public HoodieDataSourceInternalTable(String instantTime, HoodieWriteConfig config,
StructType schema, SparkSession jss, Configuration hadoopConfiguration, Map<String, String> properties,
boolean arePartitionRecordsSorted) {
boolean populateMetaFields, boolean arePartitionRecordsSorted) {
this.instantTime = instantTime;
this.writeConfig = config;
this.structType = schema;
this.jss = jss;
this.hadoopConfiguration = hadoopConfiguration;
this.arePartitionRecordsSorted = arePartitionRecordsSorted;
this.properties = properties;
this.populateMetaFields = populateMetaFields;
this.arePartitionRecordsSorted = arePartitionRecordsSorted;
}
@Override
@@ -69,7 +71,8 @@ class HoodieDataSourceInternalTable implements SupportsWrite {
@Override
public Set<TableCapability> capabilities() {
return new HashSet<TableCapability>() {{
return new HashSet<TableCapability>() {
{
add(TableCapability.BATCH_WRITE);
add(TableCapability.TRUNCATE);
}
@@ -79,6 +82,6 @@ class HoodieDataSourceInternalTable implements SupportsWrite {
@Override
public WriteBuilder newWriteBuilder(LogicalWriteInfo logicalWriteInfo) {
return new HoodieDataSourceInternalBatchWriteBuilder(instantTime, writeConfig, structType, jss,
hadoopConfiguration, properties, arePartitionRecordsSorted);
hadoopConfiguration, properties, populateMetaFields, arePartitionRecordsSorted);
}
}

28
hudi-spark-datasource/hudi-spark3/src/test/java/org/apache/hudi/spark3/internal/TestHoodieBulkInsertDataInternalWriter.java
View File

@@ -39,7 +39,6 @@ import java.util.stream.Stream;
import static org.apache.hudi.testutils.SparkDatasetTestUtils.ENCODER;
import static org.apache.hudi.testutils.SparkDatasetTestUtils.STRUCT_TYPE;
import static org.apache.hudi.testutils.SparkDatasetTestUtils.getConfigBuilder;
import static org.apache.hudi.testutils.SparkDatasetTestUtils.getInternalRowWithError;
import static org.apache.hudi.testutils.SparkDatasetTestUtils.getRandomRows;
import static org.apache.hudi.testutils.SparkDatasetTestUtils.toInternalRows;
@@ -52,6 +51,16 @@ public class TestHoodieBulkInsertDataInternalWriter extends
HoodieBulkInsertInternalWriterTestBase {
private static Stream<Arguments> configParams() {
Object[][] data = new Object[][] {
{true, true},
{true, false},
{false, true},
{false, false}
};
return Stream.of(data).map(Arguments::of);
}
private static Stream<Arguments> bulkInsertTypeParams() {
Object[][] data = new Object[][] {
{true},
{false}
@@ -61,16 +70,16 @@ public class TestHoodieBulkInsertDataInternalWriter extends
@ParameterizedTest
@MethodSource("configParams")
public void testDataInternalWriter(boolean sorted) throws Exception {
public void testDataInternalWriter(boolean sorted, boolean populateMetaFields) throws Exception {
// init config and table
HoodieWriteConfig cfg = getConfigBuilder(basePath).build();
HoodieWriteConfig cfg = getWriteConfig(populateMetaFields);
HoodieTable table = HoodieSparkTable.create(cfg, context, metaClient);
// execute N rounds
for (int i = 0; i < 5; i++) {
String instantTime = "00" + i;
// init writer
HoodieBulkInsertDataInternalWriter writer = new HoodieBulkInsertDataInternalWriter(table, cfg, instantTime, RANDOM.nextInt(100000), RANDOM.nextLong(), STRUCT_TYPE,
sorted);
HoodieBulkInsertDataInternalWriter writer = new HoodieBulkInsertDataInternalWriter(table, cfg, instantTime, RANDOM.nextInt(100000),
RANDOM.nextLong(), STRUCT_TYPE, populateMetaFields, sorted);
int size = 10 + RANDOM.nextInt(1000);
// write N rows to partition1, N rows to partition2 and N rows to partition3 ... Each batch should create a new RowCreateHandle and a new file
@@ -97,7 +106,7 @@ public class TestHoodieBulkInsertDataInternalWriter extends
// verify rows
Dataset<Row> result = sqlContext.read().parquet(fileAbsPaths.get().toArray(new String[0]));
assertOutput(totalInputRows, result, instantTime, fileNames);
assertOutput(totalInputRows, result, instantTime, fileNames, populateMetaFields);
}
}
@@ -109,12 +118,13 @@ public class TestHoodieBulkInsertDataInternalWriter extends
@Test
public void testGlobalFailure() throws Exception {
// init config and table
HoodieWriteConfig cfg = getConfigBuilder(basePath).build();
HoodieWriteConfig cfg = getWriteConfig(true);
HoodieTable table = HoodieSparkTable.create(cfg, context, metaClient);
String partitionPath = HoodieTestDataGenerator.DEFAULT_PARTITION_PATHS[0];
String instantTime = "001";
HoodieBulkInsertDataInternalWriter writer = new HoodieBulkInsertDataInternalWriter(table, cfg, instantTime, RANDOM.nextInt(100000), RANDOM.nextLong(), STRUCT_TYPE, false);
HoodieBulkInsertDataInternalWriter writer = new HoodieBulkInsertDataInternalWriter(table, cfg, instantTime, RANDOM.nextInt(100000),
RANDOM.nextLong(), STRUCT_TYPE, true, false);
int size = 10 + RANDOM.nextInt(100);
int totalFailures = 5;
@@ -150,7 +160,7 @@ public class TestHoodieBulkInsertDataInternalWriter extends
// verify rows
Dataset<Row> result = sqlContext.read().parquet(fileAbsPaths.get().toArray(new String[0]));
assertOutput(inputRows, result, instantTime, fileNames);
assertOutput(inputRows, result, instantTime, fileNames, true);
}
private void writeRows(Dataset<Row> inputRows, HoodieBulkInsertDataInternalWriter writer)

74
hudi-spark-datasource/hudi-spark3/src/test/java/org/apache/hudi/spark3/internal/TestHoodieDataSourceInternalBatchWrite.java
View File

@@ -33,6 +33,9 @@ import org.apache.spark.sql.Row;
import org.apache.spark.sql.catalyst.InternalRow;
import org.apache.spark.sql.connector.write.DataWriter;
import org.junit.jupiter.api.Test;
import org.junit.jupiter.params.ParameterizedTest;
import org.junit.jupiter.params.provider.Arguments;
import org.junit.jupiter.params.provider.MethodSource;
import java.util.ArrayList;
import java.util.Arrays;
@@ -40,10 +43,10 @@ import java.util.Collections;
import java.util.HashMap;
import java.util.List;
import java.util.Map;
import java.util.stream.Stream;
import static org.apache.hudi.testutils.SparkDatasetTestUtils.ENCODER;
import static org.apache.hudi.testutils.SparkDatasetTestUtils.STRUCT_TYPE;
import static org.apache.hudi.testutils.SparkDatasetTestUtils.getConfigBuilder;
import static org.apache.hudi.testutils.SparkDatasetTestUtils.getRandomRows;
import static org.apache.hudi.testutils.SparkDatasetTestUtils.toInternalRows;
import static org.junit.jupiter.api.Assertions.assertEquals;
@@ -55,19 +58,28 @@ import static org.junit.jupiter.api.Assertions.assertTrue;
public class TestHoodieDataSourceInternalBatchWrite extends
HoodieBulkInsertInternalWriterTestBase {
@Test
public void testDataSourceWriter() throws Exception {
testDataSourceWriterInternal(Collections.EMPTY_MAP, Collections.EMPTY_MAP);
private static Stream<Arguments> bulkInsertTypeParams() {
Object[][] data = new Object[][] {
{true},
{false}
};
return Stream.of(data).map(Arguments::of);
}
private void testDataSourceWriterInternal(Map<String, String> extraMetadata, Map<String, String> expectedExtraMetadata) throws Exception {
@ParameterizedTest
@MethodSource("bulkInsertTypeParams")
public void testDataSourceWriter(boolean populateMetaFields) throws Exception {
testDataSourceWriterInternal(Collections.EMPTY_MAP, Collections.EMPTY_MAP, populateMetaFields);
}
private void testDataSourceWriterInternal(Map<String, String> extraMetadata, Map<String, String> expectedExtraMetadata, boolean populateMetaFields) throws Exception {
// init config and table
HoodieWriteConfig cfg = getConfigBuilder(basePath).build();
HoodieWriteConfig cfg = getWriteConfig(populateMetaFields);
HoodieTable table = HoodieSparkTable.create(cfg, context, metaClient);
String instantTime = "001";
// init writer
HoodieDataSourceInternalBatchWrite dataSourceInternalBatchWrite =
new HoodieDataSourceInternalBatchWrite(instantTime, cfg, STRUCT_TYPE, sqlContext.sparkSession(), hadoopConf, extraMetadata, false);
new HoodieDataSourceInternalBatchWrite(instantTime, cfg, STRUCT_TYPE, sqlContext.sparkSession(), hadoopConf, extraMetadata, populateMetaFields, false);
DataWriter<InternalRow> writer = dataSourceInternalBatchWrite.createBatchWriterFactory(null).createWriter(0, RANDOM.nextLong());
String[] partitionPaths = HoodieTestDataGenerator.DEFAULT_PARTITION_PATHS;
@@ -99,7 +111,7 @@ public class TestHoodieDataSourceInternalBatchWrite extends
metaClient.reloadActiveTimeline();
Dataset<Row> result = HoodieClientTestUtils.read(jsc, basePath, sqlContext, metaClient.getFs(), partitionPathsAbs.toArray(new String[0]));
// verify output
assertOutput(totalInputRows, result, instantTime, Option.empty());
assertOutput(totalInputRows, result, instantTime, Option.empty(), populateMetaFields);
assertWriteStatuses(commitMessages.get(0).getWriteStatuses(), batches, size, Option.empty(), Option.empty());
// verify extra metadata
@@ -125,7 +137,7 @@ public class TestHoodieDataSourceInternalBatchWrite extends
expectedMetadata.remove(DataSourceWriteOptions.COMMIT_METADATA_KEYPREFIX_OPT_KEY().key());
expectedMetadata.remove("commit_extra_c");
testDataSourceWriterInternal(extraMeta, expectedMetadata);
testDataSourceWriterInternal(extraMeta, expectedMetadata, true);
}
@Test
@@ -137,13 +149,14 @@ public class TestHoodieDataSourceInternalBatchWrite extends
extraMeta.put("keyB", "valB");
extraMeta.put("commit_extra_c", "valC");
// none of the keys has commit metadata key prefix.
testDataSourceWriterInternal(extraMeta, Collections.EMPTY_MAP);
testDataSourceWriterInternal(extraMeta, Collections.EMPTY_MAP, true);
}
@Test
public void testMultipleDataSourceWrites() throws Exception {
@ParameterizedTest
@MethodSource("bulkInsertTypeParams")
public void testMultipleDataSourceWrites(boolean populateMetaFields) throws Exception {
// init config and table
HoodieWriteConfig cfg = getConfigBuilder(basePath).build();
HoodieWriteConfig cfg = getWriteConfig(populateMetaFields);
HoodieTable table = HoodieSparkTable.create(cfg, context, metaClient);
int partitionCounter = 0;
@@ -152,7 +165,7 @@ public class TestHoodieDataSourceInternalBatchWrite extends
String instantTime = "00" + i;
// init writer
HoodieDataSourceInternalBatchWrite dataSourceInternalBatchWrite =
new HoodieDataSourceInternalBatchWrite(instantTime, cfg, STRUCT_TYPE, sqlContext.sparkSession(), hadoopConf, Collections.EMPTY_MAP, false);
new HoodieDataSourceInternalBatchWrite(instantTime, cfg, STRUCT_TYPE, sqlContext.sparkSession(), hadoopConf, Collections.EMPTY_MAP, populateMetaFields, false);
List<HoodieWriterCommitMessage> commitMessages = new ArrayList<>();
Dataset<Row> totalInputRows = null;
DataWriter<InternalRow> writer = dataSourceInternalBatchWrite.createBatchWriterFactory(null).createWriter(partitionCounter++, RANDOM.nextLong());
@@ -176,18 +189,19 @@ public class TestHoodieDataSourceInternalBatchWrite extends
dataSourceInternalBatchWrite.commit(commitMessages.toArray(new HoodieWriterCommitMessage[0]));
metaClient.reloadActiveTimeline();
Dataset<Row> result = HoodieClientTestUtils.readCommit(basePath, sqlContext, metaClient.getCommitTimeline(), instantTime);
Dataset<Row> result = HoodieClientTestUtils.readCommit(basePath, sqlContext, metaClient.getCommitTimeline(), instantTime, populateMetaFields);
// verify output
assertOutput(totalInputRows, result, instantTime, Option.empty());
assertOutput(totalInputRows, result, instantTime, Option.empty(), populateMetaFields);
assertWriteStatuses(commitMessages.get(0).getWriteStatuses(), batches, size, Option.empty(), Option.empty());
}
}
@Test
public void testLargeWrites() throws Exception {
@ParameterizedTest
@MethodSource("bulkInsertTypeParams")
public void testLargeWrites(boolean populateMetaFields) throws Exception {
// init config and table
HoodieWriteConfig cfg = getConfigBuilder(basePath).build();
HoodieWriteConfig cfg = getWriteConfig(populateMetaFields);
HoodieTable table = HoodieSparkTable.create(cfg, context, metaClient);
int partitionCounter = 0;
@@ -196,7 +210,7 @@ public class TestHoodieDataSourceInternalBatchWrite extends
String instantTime = "00" + i;
// init writer
HoodieDataSourceInternalBatchWrite dataSourceInternalBatchWrite =
new HoodieDataSourceInternalBatchWrite(instantTime, cfg, STRUCT_TYPE, sqlContext.sparkSession(), hadoopConf, Collections.EMPTY_MAP, false);
new HoodieDataSourceInternalBatchWrite(instantTime, cfg, STRUCT_TYPE, sqlContext.sparkSession(), hadoopConf, Collections.EMPTY_MAP, populateMetaFields, false);
List<HoodieWriterCommitMessage> commitMessages = new ArrayList<>();
Dataset<Row> totalInputRows = null;
DataWriter<InternalRow> writer = dataSourceInternalBatchWrite.createBatchWriterFactory(null).createWriter(partitionCounter++, RANDOM.nextLong());
@@ -220,10 +234,11 @@ public class TestHoodieDataSourceInternalBatchWrite extends
dataSourceInternalBatchWrite.commit(commitMessages.toArray(new HoodieWriterCommitMessage[0]));
metaClient.reloadActiveTimeline();
Dataset<Row> result = HoodieClientTestUtils.readCommit(basePath, sqlContext, metaClient.getCommitTimeline(), instantTime);
Dataset<Row> result = HoodieClientTestUtils.readCommit(basePath, sqlContext, metaClient.getCommitTimeline(), instantTime,
populateMetaFields);
// verify output
assertOutput(totalInputRows, result, instantTime, Option.empty());
assertOutput(totalInputRows, result, instantTime, Option.empty(), populateMetaFields);
assertWriteStatuses(commitMessages.get(0).getWriteStatuses(), batches, size, Option.empty(), Option.empty());
}
}
@@ -234,15 +249,16 @@ public class TestHoodieDataSourceInternalBatchWrite extends
* abort batch2
* verify only records from batch1 is available to read
*/
@Test
public void testAbort() throws Exception {
@ParameterizedTest
@MethodSource("bulkInsertTypeParams")
public void testAbort(boolean populateMetaFields) throws Exception {
// init config and table
HoodieWriteConfig cfg = getConfigBuilder(basePath).build();
HoodieWriteConfig cfg = getWriteConfig(populateMetaFields);
HoodieTable table = HoodieSparkTable.create(cfg, context, metaClient);
String instantTime0 = "00" + 0;
// init writer
HoodieDataSourceInternalBatchWrite dataSourceInternalBatchWrite =
new HoodieDataSourceInternalBatchWrite(instantTime0, cfg, STRUCT_TYPE, sqlContext.sparkSession(), hadoopConf, Collections.EMPTY_MAP, false);
new HoodieDataSourceInternalBatchWrite(instantTime0, cfg, STRUCT_TYPE, sqlContext.sparkSession(), hadoopConf, Collections.EMPTY_MAP, populateMetaFields, false);
DataWriter<InternalRow> writer = dataSourceInternalBatchWrite.createBatchWriterFactory(null).createWriter(0, RANDOM.nextLong());
List<String> partitionPaths = Arrays.asList(HoodieTestDataGenerator.DEFAULT_PARTITION_PATHS);
@@ -274,13 +290,13 @@ public class TestHoodieDataSourceInternalBatchWrite extends
metaClient.reloadActiveTimeline();
Dataset<Row> result = HoodieClientTestUtils.read(jsc, basePath, sqlContext, metaClient.getFs(), partitionPathsAbs.toArray(new String[0]));
// verify rows
assertOutput(totalInputRows, result, instantTime0, Option.empty());
assertOutput(totalInputRows, result, instantTime0, Option.empty(), populateMetaFields);
assertWriteStatuses(commitMessages.get(0).getWriteStatuses(), batches, size, Option.empty(), Option.empty());
// 2nd batch. abort in the end
String instantTime1 = "00" + 1;
dataSourceInternalBatchWrite =
new HoodieDataSourceInternalBatchWrite(instantTime1, cfg, STRUCT_TYPE, sqlContext.sparkSession(), hadoopConf, Collections.EMPTY_MAP, false);
new HoodieDataSourceInternalBatchWrite(instantTime1, cfg, STRUCT_TYPE, sqlContext.sparkSession(), hadoopConf, Collections.EMPTY_MAP, populateMetaFields, false);
writer = dataSourceInternalBatchWrite.createBatchWriterFactory(null).createWriter(1, RANDOM.nextLong());
for (int j = 0; j < batches; j++) {
@@ -298,7 +314,7 @@ public class TestHoodieDataSourceInternalBatchWrite extends
result = HoodieClientTestUtils.read(jsc, basePath, sqlContext, metaClient.getFs(), partitionPathsAbs.toArray(new String[0]));
// verify rows
// only rows from first batch should be present
assertOutput(totalInputRows, result, instantTime0, Option.empty());
assertOutput(totalInputRows, result, instantTime0, Option.empty(), populateMetaFields);
}
private void writeRows(Dataset<Row> inputRows, DataWriter<InternalRow> writer) throws Exception {