Skip to content

[IcebergIO] Let hash distribution mode spill over if file gets too large #38930

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
ahmedabu98 wants to merge 3 commits into
base: master
Choose a base branch
from
+85 −57
Open

[IcebergIO] Let hash distribution mode spill over if file gets too large #38930

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
214a110
use recordwritermanager to spill over when file size gets too large
ahmedabu98 Jun 11, 2026
f627cb1
cleanup unused code
ahmedabu98 Jun 11, 2026
5f49c88
address comments
ahmedabu98 Jun 11, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
90 changes: 33 additions & 57 deletions .../io/iceberg/src/main/java/org/apache/beam/sdk/io/iceberg/WritePartitionedRowsToFiles.java
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -18,37 +18,35 @@
package org.apache.beam.sdk.io.iceberg;

import static org.apache.beam.sdk.io.iceberg.AssignDestinationsAndPartitions.DESTINATION;
import static org.apache.beam.sdk.io.iceberg.AssignDestinationsAndPartitions.PARTITION;
import static org.apache.beam.sdk.io.iceberg.RecordWriterManager.getPartitionDataPath;
import static org.apache.beam.sdk.util.Preconditions.checkStateNotNull;

import java.util.List;
import java.util.Map;
import java.util.UUID;
import org.apache.beam.sdk.coders.IterableCoder;
import org.apache.beam.sdk.coders.KvCoder;
import org.apache.beam.sdk.coders.RowCoder;
import org.apache.beam.sdk.schemas.Schema;
import org.apache.beam.sdk.transforms.DoFn;
import org.apache.beam.sdk.transforms.PTransform;
import org.apache.beam.sdk.transforms.ParDo;
import org.apache.beam.sdk.transforms.windowing.BoundedWindow;
import org.apache.beam.sdk.transforms.windowing.PaneInfo;
import org.apache.beam.sdk.values.KV;
import org.apache.beam.sdk.values.PCollection;
import org.apache.beam.sdk.values.Row;
import org.apache.beam.sdk.values.WindowedValue;
import org.apache.beam.sdk.values.WindowedValues;
import org.apache.beam.vendor.guava.v32_1_2_jre.com.google.common.collect.Maps;
import org.apache.iceberg.DataFiles;
import org.apache.iceberg.PartitionField;
import org.apache.iceberg.PartitionKey;
import org.apache.iceberg.PartitionSpec;
import org.apache.iceberg.StructLike;
import org.apache.iceberg.Table;
import org.apache.iceberg.TableProperties;
import org.apache.iceberg.catalog.Catalog;
import org.apache.iceberg.catalog.Namespace;
import org.apache.iceberg.catalog.SupportsNamespaces;
import org.apache.iceberg.catalog.TableIdentifier;
import org.apache.iceberg.data.Record;
import org.apache.iceberg.exceptions.AlreadyExistsException;
import org.apache.iceberg.exceptions.NoSuchTableException;
import org.checkerframework.checker.nullness.qual.MonotonicNonNull;
import org.apache.iceberg.util.PropertyUtil;
import org.checkerframework.checker.nullness.qual.Nullable;
import org.slf4j.Logger;
import org.slf4j.LoggerFactory;
Expand Down Expand Up @@ -87,9 +85,6 @@ private static class WriteDoFn extends DoFn<KV<Row, Iterable<Row>>, FileWriteRes
private final IcebergCatalogConfig catalogConfig;
private final String filePrefix;
private final Schema dataSchema;
private transient @MonotonicNonNull Map<TableIdentifier, Integer> specIds;
private transient @MonotonicNonNull Map<TableIdentifier, Map<String, PartitionField>>
partitionFieldMaps;

WriteDoFn(
IcebergCatalogConfig catalogConfig,
Expand All @@ -102,67 +97,48 @@ private static class WriteDoFn extends DoFn<KV<Row, Iterable<Row>>, FileWriteRes
this.dataSchema = dataSchema;
}

@Setup
public void setup() {
partitionFieldMaps = Maps.newHashMap();
specIds = Maps.newHashMap();
}

@ProcessElement
public void processElement(
@Element KV<Row, Iterable<Row>> element, OutputReceiver<FileWriteResult> out)
@Element KV<Row, Iterable<Row>> element,
BoundedWindow window,
PaneInfo paneInfo,
OutputReceiver<FileWriteResult> out)
throws Exception {
String tableIdentifier = checkStateNotNull(element.getKey().getString(DESTINATION));
String partitionPath = checkStateNotNull(element.getKey().getString(PARTITION));

IcebergDestination destination = dynamicDestinations.instantiateDestination(tableIdentifier);
Table table = getOrCreateTable(destination, dataSchema);
partitionPath =
getPartitionDataPath(
partitionPath, getPartitionFieldMap(destination.getTableIdentifier(), table));

StructLike partitionData =
table.spec().isPartitioned()
? DataFiles.data(table.spec(), partitionPath)
: new PartitionKey(table.spec(), table.schema());

String fileName =
destination
.getFileFormat()
.addExtension(String.format("%s-%s", filePrefix, UUID.randomUUID()));

RecordWriter writer =
new RecordWriter(table, destination.getFileFormat(), fileName, partitionData);

long maxFileSize =
PropertyUtil.propertyAsLong(
table.properties(),
TableProperties.WRITE_TARGET_FILE_SIZE_BYTES,
TableProperties.WRITE_TARGET_FILE_SIZE_BYTES_DEFAULT);
WindowedValue<IcebergDestination> windowedDestination =
WindowedValues.of(destination, window.maxTimestamp(), window, paneInfo);
RecordWriterManager writer =
new RecordWriterManager(catalogConfig, filePrefix, maxFileSize, Integer.MAX_VALUE);
try {
for (Row row : element.getValue()) {
Record record = IcebergUtils.beamRowToIcebergRecord(table.schema(), row);
writer.write(record);
writer.write(windowedDestination, row);
}
} finally {
writer.close();
}
Comment thread
ahmedabu98 marked this conversation as resolved.

SerializableDataFile sdf = SerializableDataFile.from(writer.getDataFile(), partitionPath);
out.output(
FileWriteResult.builder()
.setTableIdentifier(destination.getTableIdentifier())
.setSerializableDataFile(sdf)
.build());
}

private Map<String, PartitionField> getPartitionFieldMap(
TableIdentifier identifier, Table table) {
@Nullable Integer specId = checkStateNotNull(specIds).get(identifier);
if (specId != null && specId == table.spec().specId()) {
return checkStateNotNull(checkStateNotNull(partitionFieldMaps).get(identifier));
@Nullable
List<SerializableDataFile> serializableDataFiles =
writer.getSerializableDataFiles().get(windowedDestination);
if (serializableDataFiles == null) {
return;
}
Map<String, PartitionField> partitionFieldMap = Maps.newHashMap();
for (PartitionField partitionField : table.spec().fields()) {
partitionFieldMap.put(partitionField.name(), partitionField);
for (SerializableDataFile dataFile : serializableDataFiles) {
out.output(
FileWriteResult.builder()
.setTableIdentifier(destination.getTableIdentifier())
.setSerializableDataFile(dataFile)
.build());
}
Comment thread
ahmedabu98 marked this conversation as resolved.
checkStateNotNull(specIds).put(identifier, table.spec().specId());
checkStateNotNull(partitionFieldMaps).put(identifier, partitionFieldMap);
return partitionFieldMap;
}

Table getOrCreateTable(IcebergDestination destination, Schema dataSchema) {
Expand Down
52 changes: 52 additions & 0 deletions ...src/test/java/org/apache/beam/sdk/io/iceberg/IcebergWriteSchemaTransformProviderTest.java
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -58,9 +58,11 @@
import org.apache.beam.vendor.guava.v32_1_2_jre.com.google.common.collect.ImmutableList;
import org.apache.beam.vendor.guava.v32_1_2_jre.com.google.common.collect.ImmutableMap;
import org.apache.iceberg.CatalogUtil;
import org.apache.iceberg.DataFile;
import org.apache.iceberg.DistributionMode;
import org.apache.iceberg.PartitionSpec;
import org.apache.iceberg.Table;
import org.apache.iceberg.TableProperties;
import org.apache.iceberg.catalog.TableIdentifier;
import org.apache.iceberg.data.IcebergGenerics;
import org.apache.iceberg.data.Record;
Expand Down Expand Up @@ -491,6 +493,56 @@ public void writePartitionedData(boolean autosharding) {
p.run();
}

@Test
public void testWritingPartitionedDataRespectsFileSizePropertyWithSpillOver() {
Schema schema = Schema.builder().addStringField("str").addInt32Field("int").build();
org.apache.iceberg.Schema icebergSchema = IcebergUtils.beamSchemaToIcebergSchema(schema);
PartitionSpec spec = PartitionSpec.builderFor(icebergSchema).identity("str").build();
String identifier = "default.table_" + Long.toString(UUID.randomUUID().hashCode(), 16);

Table table =
warehouse.createTable(
TableIdentifier.parse(identifier),
icebergSchema,
spec,
ImmutableMap.of(TableProperties.WRITE_TARGET_FILE_SIZE_BYTES, "1"));
Map<String, Object> config =
ImmutableMap.of(
"table",
identifier,
"catalog_properties",
ImmutableMap.of("type", "hadoop", "warehouse", warehouse.location),
"distribution_mode",
distributionMode.name());

List<Row> rows =
Arrays.asList(
Row.withSchema(schema).addValues("a", 1).build(),
Row.withSchema(schema).addValues("a", 2).build(),
Row.withSchema(schema).addValues("a", 3).build(),
Row.withSchema(schema).addValues("b", 1).build(),
Row.withSchema(schema).addValues("b", 2).build(),
Row.withSchema(schema).addValues("b", 3).build());

testPipeline
.apply("Records To Add", Create.of(rows))
.setRowSchema(schema)
.apply(Managed.write(Managed.ICEBERG).withConfig(config));
testPipeline.run().waitUntilFinish();

Pipeline p = Pipeline.create(TestPipeline.testingPipelineOptions());
PCollection<Row> readRows =
p.apply(Managed.read(Managed.ICEBERG).withConfig(config)).getSinglePCollection();
PAssert.that(readRows).containsInAnyOrder(rows);

table.refresh();
List<DataFile> datafiles = new ArrayList<>();
table.snapshots().forEach(s -> s.addedDataFiles(table.io()).forEach(datafiles::add));
assertEquals(6, datafiles.size());

p.run();
}

@Test
public void testWriteCreateTableWithPartitionSpec() {
String identifier = "default.table_" + Long.toString(UUID.randomUUID().hashCode(), 16);
Expand Down
Loading