lanyuanxiaoyao/hudi
1
0
Fork 0
Code Activity

[HUDI-1147] Modify GenericRecordFullPayloadGenerator to generate vali… (#2045)

Browse Source 
* [HUDI-1147] Modify GenericRecordFullPayloadGenerator to generate valid timestamps

Co-authored-by: Sivabalan Narayanan <sivabala@uber.com>
This commit is contained in:
Balajee Nagasubramaniam
2020-12-29 13:33:19 -08:00
committed by GitHub
parent da51aa64fc
commit e33a8f733c
4 changed files with 83 additions and 22 deletions
Download Patch File Download Diff File Expand all files Collapse all files

4
hudi-integ-test/src/main/java/org/apache/hudi/integ/testsuite/generator/DeltaGenerator.java
View File

@@ -129,7 +129,7 @@ public class DeltaGenerator implements Serializable {
public JavaRDD<GenericRecord> generateInserts(Config operation) {
int numPartitions = operation.getNumInsertPartitions();
long recordsPerPartition = operation.getNumRecordsInsert() / numPartitions;
long recordsPerPartition = operation.getNumRecordsInsert();
int minPayloadSize = operation.getRecordSize();
int startPartition = operation.getStartPartition();
@@ -140,7 +140,7 @@ public class DeltaGenerator implements Serializable {
JavaRDD<GenericRecord> inputBatch = jsc.parallelize(partitionIndexes, numPartitions)
.mapPartitionsWithIndex((index, p) -> {
return new LazyRecordGeneratorIterator(new FlexibleSchemaRecordGenerationIterator(recordsPerPartition,
minPayloadSize, schemaStr, partitionPathFieldNames, (Integer)index));
minPayloadSize, schemaStr, partitionPathFieldNames, numPartitions));
}, true);
if (deltaOutputConfig.getInputParallelism() < numPartitions) {

23
hudi-integ-test/src/main/java/org/apache/hudi/integ/testsuite/generator/FlexibleSchemaRecordGenerationIterator.java
View File

@@ -20,11 +20,6 @@ package org.apache.hudi.integ.testsuite.generator;
import org.apache.avro.Schema;
import org.apache.avro.generic.GenericRecord;
import org.slf4j.Logger;
import org.slf4j.LoggerFactory;
import java.util.Iterator;
import java.util.List;
import java.util.HashSet;
import java.util.Iterator;
@@ -46,17 +41,21 @@ public class FlexibleSchemaRecordGenerationIterator implements Iterator<GenericR
private GenericRecord lastRecord;
// Partition path field name
private Set<String> partitionPathFieldNames;
private String firstPartitionPathField;
public FlexibleSchemaRecordGenerationIterator(long maxEntriesToProduce, String schema) {
this(maxEntriesToProduce, GenericRecordFullPayloadGenerator.DEFAULT_PAYLOAD_SIZE, schema, null, 0);
}
public FlexibleSchemaRecordGenerationIterator(long maxEntriesToProduce, int minPayloadSize, String schemaStr,
List<String> partitionPathFieldNames, int partitionIndex) {
List<String> partitionPathFieldNames, int numPartitions) {
this.counter = maxEntriesToProduce;
this.partitionPathFieldNames = new HashSet<>(partitionPathFieldNames);
if(partitionPathFieldNames != null && partitionPathFieldNames.size() > 0) {
this.firstPartitionPathField = partitionPathFieldNames.get(0);
}
Schema schema = new Schema.Parser().parse(schemaStr);
this.generator = new GenericRecordFullPayloadGenerator(schema, minPayloadSize, partitionIndex);
this.generator = new GenericRecordFullPayloadGenerator(schema, minPayloadSize, numPartitions);
}
@Override
@@ -67,12 +66,18 @@ public class FlexibleSchemaRecordGenerationIterator implements Iterator<GenericR
@Override
public GenericRecord next() {
this.counter--;
boolean partitionPathsNonEmpty = partitionPathFieldNames != null && partitionPathFieldNames.size() > 0;
if (lastRecord == null) {
GenericRecord record = this.generator.getNewPayload(partitionPathFieldNames);
GenericRecord record = partitionPathsNonEmpty
? this.generator.getNewPayloadWithTimestamp(this.firstPartitionPathField)
: this.generator.getNewPayload();
lastRecord = record;
return record;
} else {
return this.generator.randomize(lastRecord, this.partitionPathFieldNames);
return partitionPathsNonEmpty
? this.generator.getUpdatePayloadWithTimestamp(lastRecord,
this.partitionPathFieldNames, firstPartitionPathField)
: this.generator.getUpdatePayload(lastRecord, this.partitionPathFieldNames);
}
}
}

43
hudi-integ-test/src/main/java/org/apache/hudi/integ/testsuite/generator/GenericRecordFullPayloadGenerator.java
View File

@@ -46,9 +46,9 @@ import java.util.concurrent.TimeUnit;
*/
public class GenericRecordFullPayloadGenerator implements Serializable {
private static Logger LOG = LoggerFactory.getLogger(GenericRecordFullPayloadGenerator.class);
private static final Logger LOG = LoggerFactory.getLogger(GenericRecordFullPayloadGenerator.class);
public static final int DEFAULT_PAYLOAD_SIZE = 1024 * 10; // 10 KB
public static final int DEFAULT_NUM_DATE_PARTITIONS = 50;
public static final String DEFAULT_HOODIE_IS_DELETED_COL = "_hoodie_is_deleted";
protected final Random random = new Random();
// The source schema used to generate a payload
@@ -58,10 +58,12 @@ public class GenericRecordFullPayloadGenerator implements Serializable {
// The index of partition for which records are being generated
private int partitionIndex = 0;
// The size of a full record where every field of a generic record created contains 1 random value
private final int estimatedFullPayloadSize;
private int estimatedFullPayloadSize;
// Number of extra entries to add in a complex/collection field to achieve the desired record size
Map<String, Integer> extraEntriesMap = new HashMap<>();
// The number of unique dates to create
private int numDatePartitions = DEFAULT_NUM_DATE_PARTITIONS;
// LogicalTypes in Avro 1.8.2
private static final String DECIMAL = "decimal";
private static final String UUID_NAME = "uuid";
@@ -75,6 +77,11 @@ public class GenericRecordFullPayloadGenerator implements Serializable {
this(schema, DEFAULT_PAYLOAD_SIZE);
}
public GenericRecordFullPayloadGenerator(Schema schema, int minPayloadSize, int numDatePartitions) {
this(schema, minPayloadSize);
this.numDatePartitions = numDatePartitions;
}
public GenericRecordFullPayloadGenerator(Schema schema, int minPayloadSize) {
Pair<Integer, Integer> sizeInfo = new GenericRecordFullPayloadSizeEstimator(schema)
.typeEstimateAndNumComplexFields();
@@ -83,19 +90,13 @@ public class GenericRecordFullPayloadGenerator implements Serializable {
if (estimatedFullPayloadSize < minPayloadSize) {
int numberOfComplexFields = sizeInfo.getRight();
if (numberOfComplexFields < 1) {
LOG.warn("The schema does not have any collections/complex fields. "
+ "Cannot achieve minPayloadSize => " + minPayloadSize);
LOG.warn("The schema does not have any collections/complex fields. Cannot achieve minPayloadSize : {}",
minPayloadSize);
}
determineExtraEntriesRequired(numberOfComplexFields, minPayloadSize - estimatedFullPayloadSize);
}
}
public GenericRecordFullPayloadGenerator(Schema schema, int minPayloadSize, int partitionIndex) {
this(schema, minPayloadSize);
this.partitionIndex = partitionIndex;
}
protected static boolean isPrimitive(Schema localSchema) {
if (localSchema.getType() != Type.ARRAY
&& localSchema.getType() != Type.MAP
@@ -131,6 +132,15 @@ public class GenericRecordFullPayloadGenerator implements Serializable {
return create(baseSchema, partitionPathFieldNames);
}
public GenericRecord getNewPayloadWithTimestamp(String tsFieldName) {
return updateTimestamp(create(baseSchema, null), tsFieldName);
}
public GenericRecord getUpdatePayloadWithTimestamp(GenericRecord record, Set<String> blacklistFields,
String tsFieldName) {
return updateTimestamp(randomize(record, blacklistFields), tsFieldName);
}
protected GenericRecord create(Schema schema, Set<String> partitionPathFieldNames) {
GenericRecord result = new GenericData.Record(schema);
for (Schema.Field f : schema.getFields()) {
@@ -314,6 +324,17 @@ public class GenericRecordFullPayloadGenerator implements Serializable {
return genericData.validate(baseSchema, record);
}
/*
* Generates a sequential timestamp (daily increment), and updates the timestamp field of the record.
* Note: When generating records, number of records to be generated must be more than numDatePartitions * parallelism,
* to guarantee that at least numDatePartitions are created.
*/
public GenericRecord updateTimestamp(GenericRecord record, String fieldName) {
long delta = TimeUnit.MILLISECONDS.convert(++partitionIndex % numDatePartitions, TimeUnit.DAYS);
record.put(fieldName, System.currentTimeMillis() - delta);
return record;
}
/**
* Check whether a schema is option. return true if it match the follows: 1. Its type is Type.UNION 2. Has two types 3. Has a NULL type.
*/