stevenzwu commented on code in PR #8553: URL: https://github.com/apache/iceberg/pull/8553#discussion_r1339231569
########## flink/v1.17/flink/src/main/java/org/apache/iceberg/flink/source/reader/IcebergTimestampWatermarkExtractor.java: ########## @@ -0,0 +1,65 @@ +/* + * Licensed to the Apache Software Foundation (ASF) under one + * or more contributor license agreements. See the NOTICE file + * distributed with this work for additional information + * regarding copyright ownership. The ASF licenses this file + * to you under the Apache License, Version 2.0 (the + * "License"); you may not use this file except in compliance + * with the License. You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, + * software distributed under the License is distributed on an + * "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY + * KIND, either express or implied. See the License for the + * specific language governing permissions and limitations + * under the License. + */ +package org.apache.iceberg.flink.source.reader; + +import java.io.Serializable; +import java.util.Comparator; +import org.apache.flink.table.data.RowData; +import org.apache.iceberg.Schema; +import org.apache.iceberg.flink.source.split.IcebergSourceSplit; +import org.apache.iceberg.relocated.com.google.common.base.Preconditions; +import org.apache.iceberg.types.Conversions; +import org.apache.iceberg.types.Type; +import org.apache.iceberg.types.Types; + +/** + * {@link IcebergWatermarkExtractor} implementation which uses an Iceberg timestamp column + * statistics to get the watermarks for the {@link IcebergSourceSplit}. This watermark is emitted by + * the {@link WatermarkExtractorRecordEmitter} along with the actual records. + */ +public class IcebergTimestampWatermarkExtractor Review Comment: I would keep either `Timestamp` or `Watermark`, not both. e.g. `IcebergSourceSplitWatermarkAssigner` ########## flink/v1.17/flink/src/main/java/org/apache/iceberg/flink/source/reader/IcebergWatermarkExtractor.java: ########## @@ -0,0 +1,28 @@ +/* + * Licensed to the Apache Software Foundation (ASF) under one + * or more contributor license agreements. See the NOTICE file + * distributed with this work for additional information + * regarding copyright ownership. The ASF licenses this file + * to you under the Apache License, Version 2.0 (the + * "License"); you may not use this file except in compliance + * with the License. You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, + * software distributed under the License is distributed on an + * "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY + * KIND, either express or implied. See the License for the + * specific language governing permissions and limitations + * under the License. + */ +package org.apache.iceberg.flink.source.reader; + +import java.io.Serializable; +import org.apache.iceberg.flink.source.split.IcebergSourceSplit; + +/** The interface used to extract watermarks from splits. */ +public interface IcebergWatermarkExtractor<T> extends Serializable { Review Comment: we probably can use the `TimestampAssigner<T>` interface here too. `T` could be `IcebergSourceSplit`. We can add an implementation of `IcebergSourceSplitTimestampAssigner(String fieldName, TimeUnit unit)`. I am also thinking it can be non-public. Maybe `IcebergSource$Builder` just exposes `timestampField(String fieldName, TimeUnit unit)`? ########## flink/v1.17/flink/src/test/java/org/apache/iceberg/flink/source/TestIcebergSourceWithWatermarkExtractor.java: ########## @@ -0,0 +1,377 @@ +/* + * Licensed to the Apache Software Foundation (ASF) under one + * or more contributor license agreements. See the NOTICE file + * distributed with this work for additional information + * regarding copyright ownership. The ASF licenses this file + * to you under the Apache License, Version 2.0 (the + * "License"); you may not use this file except in compliance + * with the License. You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, + * software distributed under the License is distributed on an + * "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY + * KIND, either express or implied. See the License for the + * specific language governing permissions and limitations + * under the License. + */ +package org.apache.iceberg.flink.source; + +import static org.apache.flink.connector.testframe.utils.ConnectorTestConstants.DEFAULT_COLLECT_DATA_TIMEOUT; +import static org.assertj.core.api.AssertionsForClassTypes.assertThat; + +import java.io.Serializable; +import java.time.Duration; +import java.time.Instant; +import java.time.LocalDateTime; +import java.time.ZoneId; +import java.util.List; +import java.util.Optional; +import java.util.Set; +import java.util.UUID; +import java.util.concurrent.CompletableFuture; +import java.util.concurrent.TimeUnit; +import java.util.stream.Collectors; +import org.apache.flink.api.common.JobID; +import org.apache.flink.api.common.eventtime.SerializableTimestampAssigner; +import org.apache.flink.api.common.eventtime.WatermarkStrategy; +import org.apache.flink.api.common.typeinfo.TypeInformation; +import org.apache.flink.api.common.typeutils.TypeSerializer; +import org.apache.flink.configuration.Configuration; +import org.apache.flink.core.execution.JobClient; +import org.apache.flink.metrics.Gauge; +import org.apache.flink.runtime.metrics.MetricNames; +import org.apache.flink.runtime.minicluster.RpcServiceSharing; +import org.apache.flink.runtime.testutils.InMemoryReporter; +import org.apache.flink.runtime.testutils.MiniClusterResourceConfiguration; +import org.apache.flink.streaming.api.datastream.DataStream; +import org.apache.flink.streaming.api.environment.StreamExecutionEnvironment; +import org.apache.flink.streaming.api.functions.windowing.AllWindowFunction; +import org.apache.flink.streaming.api.operators.collect.CollectResultIterator; +import org.apache.flink.streaming.api.operators.collect.CollectSinkOperator; +import org.apache.flink.streaming.api.operators.collect.CollectSinkOperatorFactory; +import org.apache.flink.streaming.api.operators.collect.CollectStreamSink; +import org.apache.flink.streaming.api.windowing.assigners.TumblingEventTimeWindows; +import org.apache.flink.streaming.api.windowing.time.Time; +import org.apache.flink.streaming.api.windowing.windows.TimeWindow; +import org.apache.flink.table.data.RowData; +import org.apache.flink.test.util.MiniClusterWithClientResource; +import org.apache.flink.util.Collector; +import org.apache.iceberg.FileFormat; +import org.apache.iceberg.data.GenericAppenderHelper; +import org.apache.iceberg.data.GenericRecord; +import org.apache.iceberg.data.RandomGenericData; +import org.apache.iceberg.data.Record; +import org.apache.iceberg.flink.HadoopTableResource; +import org.apache.iceberg.flink.RowDataConverter; +import org.apache.iceberg.flink.TestFixtures; +import org.apache.iceberg.flink.source.reader.IcebergTimestampWatermarkExtractor; +import org.apache.iceberg.relocated.com.google.common.collect.ImmutableList; +import org.apache.iceberg.relocated.com.google.common.collect.Lists; +import org.apache.iceberg.relocated.com.google.common.collect.Sets; +import org.apache.iceberg.types.Comparators; +import org.apache.iceberg.util.StructLikeWrapper; +import org.awaitility.Awaitility; +import org.junit.Assert; +import org.junit.ClassRule; +import org.junit.Rule; +import org.junit.Test; +import org.junit.rules.TemporaryFolder; + +public class TestIcebergSourceWithWatermarkExtractor implements Serializable { Review Comment: can we change to junit5? there is already `HadoopCatalogExtension` in Iceberg repo and `MiniClusterExtension` in Flink ########## flink/v1.17/flink/src/main/java/org/apache/iceberg/flink/source/reader/SerializableRecordEmitter.java: ########## @@ -0,0 +1,39 @@ +/* + * Licensed to the Apache Software Foundation (ASF) under one + * or more contributor license agreements. See the NOTICE file + * distributed with this work for additional information + * regarding copyright ownership. The ASF licenses this file + * to you under the Apache License, Version 2.0 (the + * "License"); you may not use this file except in compliance + * with the License. You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, + * software distributed under the License is distributed on an + * "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY + * KIND, either express or implied. See the License for the + * specific language governing permissions and limitations + * under the License. + */ +package org.apache.iceberg.flink.source.reader; + +import java.io.Serializable; +import org.apache.flink.connector.base.source.reader.RecordEmitter; +import org.apache.iceberg.flink.source.split.IcebergSourceSplit; + +@FunctionalInterface +public interface SerializableRecordEmitter<T> Review Comment: annotate with `@Internal`? ########## flink/v1.17/flink/src/test/java/org/apache/iceberg/flink/source/TestIcebergSourceWithWatermarkExtractor.java: ########## @@ -0,0 +1,377 @@ +/* + * Licensed to the Apache Software Foundation (ASF) under one + * or more contributor license agreements. See the NOTICE file + * distributed with this work for additional information + * regarding copyright ownership. The ASF licenses this file + * to you under the Apache License, Version 2.0 (the + * "License"); you may not use this file except in compliance + * with the License. You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, + * software distributed under the License is distributed on an + * "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY + * KIND, either express or implied. See the License for the + * specific language governing permissions and limitations + * under the License. + */ +package org.apache.iceberg.flink.source; + +import static org.apache.flink.connector.testframe.utils.ConnectorTestConstants.DEFAULT_COLLECT_DATA_TIMEOUT; +import static org.assertj.core.api.AssertionsForClassTypes.assertThat; + +import java.io.Serializable; +import java.time.Duration; +import java.time.Instant; +import java.time.LocalDateTime; +import java.time.ZoneId; +import java.util.List; +import java.util.Optional; +import java.util.Set; +import java.util.UUID; +import java.util.concurrent.CompletableFuture; +import java.util.concurrent.TimeUnit; +import java.util.stream.Collectors; +import org.apache.flink.api.common.JobID; +import org.apache.flink.api.common.eventtime.SerializableTimestampAssigner; +import org.apache.flink.api.common.eventtime.WatermarkStrategy; +import org.apache.flink.api.common.typeinfo.TypeInformation; +import org.apache.flink.api.common.typeutils.TypeSerializer; +import org.apache.flink.configuration.Configuration; +import org.apache.flink.core.execution.JobClient; +import org.apache.flink.metrics.Gauge; +import org.apache.flink.runtime.metrics.MetricNames; +import org.apache.flink.runtime.minicluster.RpcServiceSharing; +import org.apache.flink.runtime.testutils.InMemoryReporter; +import org.apache.flink.runtime.testutils.MiniClusterResourceConfiguration; +import org.apache.flink.streaming.api.datastream.DataStream; +import org.apache.flink.streaming.api.environment.StreamExecutionEnvironment; +import org.apache.flink.streaming.api.functions.windowing.AllWindowFunction; +import org.apache.flink.streaming.api.operators.collect.CollectResultIterator; +import org.apache.flink.streaming.api.operators.collect.CollectSinkOperator; +import org.apache.flink.streaming.api.operators.collect.CollectSinkOperatorFactory; +import org.apache.flink.streaming.api.operators.collect.CollectStreamSink; +import org.apache.flink.streaming.api.windowing.assigners.TumblingEventTimeWindows; +import org.apache.flink.streaming.api.windowing.time.Time; +import org.apache.flink.streaming.api.windowing.windows.TimeWindow; +import org.apache.flink.table.data.RowData; +import org.apache.flink.test.util.MiniClusterWithClientResource; +import org.apache.flink.util.Collector; +import org.apache.iceberg.FileFormat; +import org.apache.iceberg.data.GenericAppenderHelper; +import org.apache.iceberg.data.GenericRecord; +import org.apache.iceberg.data.RandomGenericData; +import org.apache.iceberg.data.Record; +import org.apache.iceberg.flink.HadoopTableResource; +import org.apache.iceberg.flink.RowDataConverter; +import org.apache.iceberg.flink.TestFixtures; +import org.apache.iceberg.flink.source.reader.IcebergTimestampWatermarkExtractor; +import org.apache.iceberg.relocated.com.google.common.collect.ImmutableList; +import org.apache.iceberg.relocated.com.google.common.collect.Lists; +import org.apache.iceberg.relocated.com.google.common.collect.Sets; +import org.apache.iceberg.types.Comparators; +import org.apache.iceberg.util.StructLikeWrapper; +import org.awaitility.Awaitility; +import org.junit.Assert; +import org.junit.ClassRule; +import org.junit.Rule; +import org.junit.Test; +import org.junit.rules.TemporaryFolder; + +public class TestIcebergSourceWithWatermarkExtractor implements Serializable { + private static final InMemoryReporter reporter = InMemoryReporter.createWithRetainedMetrics(); + private static final int PARALLELISM = 4; + private static final String SOURCE_NAME = "IcebergSource"; + private static final int RECORD_NUM_FOR_2_SPLITS = 200; + + @ClassRule public static final TemporaryFolder TEMPORARY_FOLDER = new TemporaryFolder(); + + @Rule + public final MiniClusterWithClientResource miniClusterResource = + new MiniClusterWithClientResource( + new MiniClusterResourceConfiguration.Builder() + .setNumberTaskManagers(1) + .setNumberSlotsPerTaskManager(PARALLELISM) + .setRpcServiceSharing(RpcServiceSharing.DEDICATED) + .setConfiguration(reporter.addToConfiguration(new Configuration())) + .withHaLeadershipControl() + .build()); + + @Rule + public final HadoopTableResource sourceTableResource = + new HadoopTableResource( + TEMPORARY_FOLDER, TestFixtures.DATABASE, TestFixtures.TABLE, TestFixtures.TS_SCHEMA); + + @Test + public void testWindowing() throws Exception { + GenericAppenderHelper dataAppender = appender(); + List<Record> expectedRecords = Lists.newArrayList(); + + // Generate records with the following pattern: + // - File 1 - Later records (Watermark 6000000) Review Comment: don't quite understand the watermark value here. Doesn't the first record have timestamp of 100? ########## flink/v1.17/flink/src/test/java/org/apache/iceberg/flink/source/TestIcebergSourceWithWatermarkExtractor.java: ########## @@ -0,0 +1,377 @@ +/* + * Licensed to the Apache Software Foundation (ASF) under one + * or more contributor license agreements. See the NOTICE file + * distributed with this work for additional information + * regarding copyright ownership. The ASF licenses this file + * to you under the Apache License, Version 2.0 (the + * "License"); you may not use this file except in compliance + * with the License. You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, + * software distributed under the License is distributed on an + * "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY + * KIND, either express or implied. See the License for the + * specific language governing permissions and limitations + * under the License. + */ +package org.apache.iceberg.flink.source; + +import static org.apache.flink.connector.testframe.utils.ConnectorTestConstants.DEFAULT_COLLECT_DATA_TIMEOUT; +import static org.assertj.core.api.AssertionsForClassTypes.assertThat; + +import java.io.Serializable; +import java.time.Duration; +import java.time.Instant; +import java.time.LocalDateTime; +import java.time.ZoneId; +import java.util.List; +import java.util.Optional; +import java.util.Set; +import java.util.UUID; +import java.util.concurrent.CompletableFuture; +import java.util.concurrent.TimeUnit; +import java.util.stream.Collectors; +import org.apache.flink.api.common.JobID; +import org.apache.flink.api.common.eventtime.SerializableTimestampAssigner; +import org.apache.flink.api.common.eventtime.WatermarkStrategy; +import org.apache.flink.api.common.typeinfo.TypeInformation; +import org.apache.flink.api.common.typeutils.TypeSerializer; +import org.apache.flink.configuration.Configuration; +import org.apache.flink.core.execution.JobClient; +import org.apache.flink.metrics.Gauge; +import org.apache.flink.runtime.metrics.MetricNames; +import org.apache.flink.runtime.minicluster.RpcServiceSharing; +import org.apache.flink.runtime.testutils.InMemoryReporter; +import org.apache.flink.runtime.testutils.MiniClusterResourceConfiguration; +import org.apache.flink.streaming.api.datastream.DataStream; +import org.apache.flink.streaming.api.environment.StreamExecutionEnvironment; +import org.apache.flink.streaming.api.functions.windowing.AllWindowFunction; +import org.apache.flink.streaming.api.operators.collect.CollectResultIterator; +import org.apache.flink.streaming.api.operators.collect.CollectSinkOperator; +import org.apache.flink.streaming.api.operators.collect.CollectSinkOperatorFactory; +import org.apache.flink.streaming.api.operators.collect.CollectStreamSink; +import org.apache.flink.streaming.api.windowing.assigners.TumblingEventTimeWindows; +import org.apache.flink.streaming.api.windowing.time.Time; +import org.apache.flink.streaming.api.windowing.windows.TimeWindow; +import org.apache.flink.table.data.RowData; +import org.apache.flink.test.util.MiniClusterWithClientResource; +import org.apache.flink.util.Collector; +import org.apache.iceberg.FileFormat; +import org.apache.iceberg.data.GenericAppenderHelper; +import org.apache.iceberg.data.GenericRecord; +import org.apache.iceberg.data.RandomGenericData; +import org.apache.iceberg.data.Record; +import org.apache.iceberg.flink.HadoopTableResource; +import org.apache.iceberg.flink.RowDataConverter; +import org.apache.iceberg.flink.TestFixtures; +import org.apache.iceberg.flink.source.reader.IcebergTimestampWatermarkExtractor; +import org.apache.iceberg.relocated.com.google.common.collect.ImmutableList; +import org.apache.iceberg.relocated.com.google.common.collect.Lists; +import org.apache.iceberg.relocated.com.google.common.collect.Sets; +import org.apache.iceberg.types.Comparators; +import org.apache.iceberg.util.StructLikeWrapper; +import org.awaitility.Awaitility; +import org.junit.Assert; +import org.junit.ClassRule; +import org.junit.Rule; +import org.junit.Test; +import org.junit.rules.TemporaryFolder; + +public class TestIcebergSourceWithWatermarkExtractor implements Serializable { + private static final InMemoryReporter reporter = InMemoryReporter.createWithRetainedMetrics(); + private static final int PARALLELISM = 4; + private static final String SOURCE_NAME = "IcebergSource"; + private static final int RECORD_NUM_FOR_2_SPLITS = 200; + + @ClassRule public static final TemporaryFolder TEMPORARY_FOLDER = new TemporaryFolder(); + + @Rule + public final MiniClusterWithClientResource miniClusterResource = + new MiniClusterWithClientResource( + new MiniClusterResourceConfiguration.Builder() + .setNumberTaskManagers(1) + .setNumberSlotsPerTaskManager(PARALLELISM) + .setRpcServiceSharing(RpcServiceSharing.DEDICATED) + .setConfiguration(reporter.addToConfiguration(new Configuration())) + .withHaLeadershipControl() + .build()); + + @Rule + public final HadoopTableResource sourceTableResource = + new HadoopTableResource( + TEMPORARY_FOLDER, TestFixtures.DATABASE, TestFixtures.TABLE, TestFixtures.TS_SCHEMA); + + @Test + public void testWindowing() throws Exception { + GenericAppenderHelper dataAppender = appender(); + List<Record> expectedRecords = Lists.newArrayList(); + + // Generate records with the following pattern: + // - File 1 - Later records (Watermark 6000000) + // - Split 1 - 2 records (100, "File-1-100"), (103, "File-1-103") + // - File 2 - First records (Watermark 0) Review Comment: how did we achieve 2 splits in file 2? didn't see explicit setting of small target split size? ########## flink/v1.17/flink/src/main/java/org/apache/iceberg/flink/source/reader/WatermarkExtractorRecordEmitter.java: ########## @@ -0,0 +1,64 @@ +/* + * Licensed to the Apache Software Foundation (ASF) under one + * or more contributor license agreements. See the NOTICE file + * distributed with this work for additional information + * regarding copyright ownership. The ASF licenses this file + * to you under the Apache License, Version 2.0 (the + * "License"); you may not use this file except in compliance + * with the License. You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, + * software distributed under the License is distributed on an + * "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY + * KIND, either express or implied. See the License for the + * specific language governing permissions and limitations + * under the License. + */ +package org.apache.iceberg.flink.source.reader; + +import org.apache.flink.api.common.eventtime.Watermark; +import org.apache.flink.api.connector.source.SourceOutput; +import org.apache.iceberg.flink.source.split.IcebergSourceSplit; +import org.slf4j.Logger; +import org.slf4j.LoggerFactory; + +/** + * Emitter which emits the watermarks, records and updates the split position. + * + * <p>The Emitter emits watermarks at the beginning of every split provided by the {@link + * IcebergWatermarkExtractor}. + */ +class WatermarkExtractorRecordEmitter<T> implements SerializableRecordEmitter<T> { + private static final Logger LOG = LoggerFactory.getLogger(WatermarkExtractorRecordEmitter.class); + private final IcebergWatermarkExtractor timeExtractor; + private String lastSplit = null; + private long watermark; + + WatermarkExtractorRecordEmitter(IcebergWatermarkExtractor timeExtractor) { + this.timeExtractor = timeExtractor; + } + + @Override + public void emitRecord( + RecordAndPosition<T> element, SourceOutput<T> output, IcebergSourceSplit split) { + if (!split.splitId().equals(lastSplit)) { + long newWatermark = timeExtractor.extractWatermark(split); + if (newWatermark < watermark) { + LOG.warn( + "Watermark decreased. PreviousWM {}, currentWM {}, previousSplit {}, currentSplit {}.", + watermark, + newWatermark, + lastSplit, + split.splitId()); + } + watermark = newWatermark; + output.emitWatermark(new Watermark(watermark)); + lastSplit = split.splitId(); + } + + output.collect(element.record()); + split.updatePosition(element.fileOffset(), element.recordOffset()); + } Review Comment: I am fine with this since we are talking 2 lines of code. But I am not sure if inheritance affects performance visibly in in hot path. ########## flink/v1.17/flink/src/test/java/org/apache/iceberg/flink/source/TestIcebergSourceWithWatermarkExtractor.java: ########## @@ -0,0 +1,377 @@ +/* + * Licensed to the Apache Software Foundation (ASF) under one + * or more contributor license agreements. See the NOTICE file + * distributed with this work for additional information + * regarding copyright ownership. The ASF licenses this file + * to you under the Apache License, Version 2.0 (the + * "License"); you may not use this file except in compliance + * with the License. You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, + * software distributed under the License is distributed on an + * "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY + * KIND, either express or implied. See the License for the + * specific language governing permissions and limitations + * under the License. + */ +package org.apache.iceberg.flink.source; + +import static org.apache.flink.connector.testframe.utils.ConnectorTestConstants.DEFAULT_COLLECT_DATA_TIMEOUT; +import static org.assertj.core.api.AssertionsForClassTypes.assertThat; + +import java.io.Serializable; +import java.time.Duration; +import java.time.Instant; +import java.time.LocalDateTime; +import java.time.ZoneId; +import java.util.List; +import java.util.Optional; +import java.util.Set; +import java.util.UUID; +import java.util.concurrent.CompletableFuture; +import java.util.concurrent.TimeUnit; +import java.util.stream.Collectors; +import org.apache.flink.api.common.JobID; +import org.apache.flink.api.common.eventtime.SerializableTimestampAssigner; +import org.apache.flink.api.common.eventtime.WatermarkStrategy; +import org.apache.flink.api.common.typeinfo.TypeInformation; +import org.apache.flink.api.common.typeutils.TypeSerializer; +import org.apache.flink.configuration.Configuration; +import org.apache.flink.core.execution.JobClient; +import org.apache.flink.metrics.Gauge; +import org.apache.flink.runtime.metrics.MetricNames; +import org.apache.flink.runtime.minicluster.RpcServiceSharing; +import org.apache.flink.runtime.testutils.InMemoryReporter; +import org.apache.flink.runtime.testutils.MiniClusterResourceConfiguration; +import org.apache.flink.streaming.api.datastream.DataStream; +import org.apache.flink.streaming.api.environment.StreamExecutionEnvironment; +import org.apache.flink.streaming.api.functions.windowing.AllWindowFunction; +import org.apache.flink.streaming.api.operators.collect.CollectResultIterator; +import org.apache.flink.streaming.api.operators.collect.CollectSinkOperator; +import org.apache.flink.streaming.api.operators.collect.CollectSinkOperatorFactory; +import org.apache.flink.streaming.api.operators.collect.CollectStreamSink; +import org.apache.flink.streaming.api.windowing.assigners.TumblingEventTimeWindows; +import org.apache.flink.streaming.api.windowing.time.Time; +import org.apache.flink.streaming.api.windowing.windows.TimeWindow; +import org.apache.flink.table.data.RowData; +import org.apache.flink.test.util.MiniClusterWithClientResource; +import org.apache.flink.util.Collector; +import org.apache.iceberg.FileFormat; +import org.apache.iceberg.data.GenericAppenderHelper; +import org.apache.iceberg.data.GenericRecord; +import org.apache.iceberg.data.RandomGenericData; +import org.apache.iceberg.data.Record; +import org.apache.iceberg.flink.HadoopTableResource; +import org.apache.iceberg.flink.RowDataConverter; +import org.apache.iceberg.flink.TestFixtures; +import org.apache.iceberg.flink.source.reader.IcebergTimestampWatermarkExtractor; +import org.apache.iceberg.relocated.com.google.common.collect.ImmutableList; +import org.apache.iceberg.relocated.com.google.common.collect.Lists; +import org.apache.iceberg.relocated.com.google.common.collect.Sets; +import org.apache.iceberg.types.Comparators; +import org.apache.iceberg.util.StructLikeWrapper; +import org.awaitility.Awaitility; +import org.junit.Assert; +import org.junit.ClassRule; +import org.junit.Rule; +import org.junit.Test; +import org.junit.rules.TemporaryFolder; + +public class TestIcebergSourceWithWatermarkExtractor implements Serializable { + private static final InMemoryReporter reporter = InMemoryReporter.createWithRetainedMetrics(); + private static final int PARALLELISM = 4; + private static final String SOURCE_NAME = "IcebergSource"; + private static final int RECORD_NUM_FOR_2_SPLITS = 200; + + @ClassRule public static final TemporaryFolder TEMPORARY_FOLDER = new TemporaryFolder(); + + @Rule + public final MiniClusterWithClientResource miniClusterResource = + new MiniClusterWithClientResource( + new MiniClusterResourceConfiguration.Builder() + .setNumberTaskManagers(1) + .setNumberSlotsPerTaskManager(PARALLELISM) + .setRpcServiceSharing(RpcServiceSharing.DEDICATED) + .setConfiguration(reporter.addToConfiguration(new Configuration())) + .withHaLeadershipControl() + .build()); + + @Rule + public final HadoopTableResource sourceTableResource = + new HadoopTableResource( + TEMPORARY_FOLDER, TestFixtures.DATABASE, TestFixtures.TABLE, TestFixtures.TS_SCHEMA); + + @Test + public void testWindowing() throws Exception { + GenericAppenderHelper dataAppender = appender(); + List<Record> expectedRecords = Lists.newArrayList(); + + // Generate records with the following pattern: + // - File 1 - Later records (Watermark 6000000) + // - Split 1 - 2 records (100, "File-1-100"), (103, "File-1-103") + // - File 2 - First records (Watermark 0) + // - Split 1 - 100 records (0, "File-2-0"), (1, "File-2-1"),... + // - Split 2 - 100 records (0, "File-2-0"), (1, "File-2-1"),... + // - File 3 - Parallel write for the first records (Watermark 60000) + // - Split 1 - 2 records (1, "File-3-1"), (3, "File-3-3") + List<Record> batch; + batch = ImmutableList.of(generateRecord(100, "100"), generateRecord(103, "103")); + expectedRecords.addAll(batch); + dataAppender.appendToTable(batch); + + batch = Lists.newArrayListWithCapacity(100); + for (int i = 0; i < RECORD_NUM_FOR_2_SPLITS; ++i) { + batch.add(generateRecord(i % 5, "File-2-" + i)); Review Comment: not sure what the `% 5` achieve here? ########## flink/v1.17/flink/src/main/java/org/apache/iceberg/flink/source/IcebergSource.java: ########## @@ -429,6 +440,20 @@ public Builder<T> setAll(Map<String, String> properties) { return this; } + /** + * Sets the {@link IcebergWatermarkExtractor} to retrieve the split watermark before emitting + * the records for a given split. The {@link + * IcebergWatermarkExtractor#extractWatermark(IcebergSourceSplit)} is also used for ordering the + * splits for read. + */ + public Builder<T> watermarkExtractor(IcebergWatermarkExtractor<T> newWatermarkExtractor) { Review Comment: I can see where @sundargates came from. Original API exposes `SplitAssignerFactory` in the `IcebergSource$Builder`. the intention is to allow users to switch to different assigners (like ordered, locality aware assigners). @sundargates with the new approach, readers need to extract split level watermark and emit the watermark values. Hence just passing timestamp extractor to assigner is not enough. But I also realized that maybe exposing `SplitAssignerFactory` is not very intuitive to end users. In another comment, I suggested `sourceTimestampField (or) sourceWatermarkField(String fieldName, TimeUnit unit)` ########## flink/v1.17/flink/src/test/java/org/apache/iceberg/flink/source/TestIcebergSourceFailoverWithWatermarkExtractor.java: ########## @@ -0,0 +1,116 @@ +/* + * Licensed to the Apache Software Foundation (ASF) under one + * or more contributor license agreements. See the NOTICE file + * distributed with this work for additional information + * regarding copyright ownership. The ASF licenses this file + * to you under the Apache License, Version 2.0 (the + * "License"); you may not use this file except in compliance + * with the License. You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, + * software distributed under the License is distributed on an + * "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY + * KIND, either express or implied. See the License for the + * specific language governing permissions and limitations + * under the License. + */ +package org.apache.iceberg.flink.source; + +import java.time.Duration; +import java.time.Instant; +import java.time.LocalDateTime; +import java.time.ZoneId; +import java.time.ZoneOffset; +import java.util.List; +import java.util.concurrent.TimeUnit; +import java.util.concurrent.atomic.AtomicLong; +import java.util.stream.Collectors; +import org.apache.flink.table.data.RowData; +import org.apache.iceberg.Schema; +import org.apache.iceberg.Table; +import org.apache.iceberg.data.RandomGenericData; +import org.apache.iceberg.data.Record; +import org.apache.iceberg.flink.SimpleDataUtil; +import org.apache.iceberg.flink.TestFixtures; +import org.apache.iceberg.flink.source.reader.IcebergTimestampWatermarkExtractor; +import org.apache.iceberg.types.Comparators; +import org.apache.iceberg.util.StructLikeWrapper; + +public class TestIcebergSourceFailoverWithWatermarkExtractor extends TestIcebergSourceFailover { + // Increment ts by 15 minutes for each generateRecords batch + private static final long RECORD_BATCH_TS_INCREMENT_MILLI = TimeUnit.MINUTES.toMillis(15); + // Within a batch, increment ts by 1 minute Review Comment: 1 minute -> 1 second ########## flink/v1.17/flink/src/main/java/org/apache/iceberg/flink/source/eventtimeextractor/EventTimeExtractorRecordEmitter.java: ########## @@ -0,0 +1,76 @@ +/* + * Licensed to the Apache Software Foundation (ASF) under one + * or more contributor license agreements. See the NOTICE file + * distributed with this work for additional information + * regarding copyright ownership. The ASF licenses this file + * to you under the Apache License, Version 2.0 (the + * "License"); you may not use this file except in compliance + * with the License. You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, + * software distributed under the License is distributed on an + * "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY + * KIND, either express or implied. See the License for the + * specific language governing permissions and limitations + * under the License. + */ +package org.apache.iceberg.flink.source.eventtimeextractor; + +import org.apache.flink.api.common.eventtime.Watermark; +import org.apache.flink.api.connector.source.SourceOutput; +import org.apache.iceberg.flink.source.reader.RecordAndPosition; +import org.apache.iceberg.flink.source.reader.SerializableRecordEmitter; +import org.apache.iceberg.flink.source.split.IcebergSourceSplit; +import org.slf4j.Logger; +import org.slf4j.LoggerFactory; + +/** + * Emitter which emits the record with event time and updates the split position. + * + * <p>The Emitter also emits watermarks at the beginning of every split, and sets the event + * timestamp based on the provided {@link IcebergEventTimeExtractor}. + */ +public final class EventTimeExtractorRecordEmitter<T> implements SerializableRecordEmitter<T> { + private static final Logger LOG = LoggerFactory.getLogger(EventTimeExtractorRecordEmitter.class); + private final IcebergEventTimeExtractor timeExtractor; + private String lastSplit = null; + private long watermark; + + public EventTimeExtractorRecordEmitter(IcebergEventTimeExtractor timeExtractor) { + this.timeExtractor = timeExtractor; + } + + @Override + public void emitRecord( + RecordAndPosition<T> element, SourceOutput<T> output, IcebergSourceSplit split) { + if (!split.splitId().equals(lastSplit)) { + long newWatermark = timeExtractor.extractWatermark(split); + if (newWatermark < watermark) { + LOG.warn( + "Watermark decreased. PreviousWM {}, currentWM {}, previousSplit {}, currentSplit {}.", + watermark, + newWatermark, + lastSplit, + split.splitId()); + } + watermark = newWatermark; + output.emitWatermark(new Watermark(watermark)); + lastSplit = split.splitId(); + } + + long eventTime = timeExtractor.extractEventTime(element.record()); + if (eventTime <= watermark) { + LOG.warn( + "Late event arrived. PreviousWM {}, split {}, eventTime {}, record {}.", + watermark, + split, + eventTime, + element.record()); + } + + output.collect(element.record(), eventTime); Review Comment: > it is unclear to me if the watermark strategy is passed into the source operator (like Iceberg), how does it interfere/live with the watermark emitted by the emitter at split level. @pvary this is the part I don't know. the source watermark alignment works with emitter emitting split level watermark values that should advance watermark. but then if the watermark strategy was passed into the source operator directly. how are the watermark actually working especially related to the watermark alignment in the source reader. -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected] --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
