stevenzwu commented on code in PR #10308: URL: https://github.com/apache/iceberg/pull/10308#discussion_r1608720543
########## flink/v1.19/flink/src/main/java/org/apache/iceberg/flink/maintenance/operator/SingleThreadedIteratorSource.java: ########## @@ -0,0 +1,195 @@ +/* + * 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.maintenance.operator; + +import java.io.IOException; +import java.util.Collection; +import java.util.Iterator; +import org.apache.flink.api.connector.source.Source; +import org.apache.flink.api.connector.source.SourceReader; +import org.apache.flink.api.connector.source.SourceReaderContext; +import org.apache.flink.api.connector.source.SplitEnumerator; +import org.apache.flink.api.connector.source.SplitEnumeratorContext; +import org.apache.flink.api.connector.source.lib.util.IteratorSourceEnumerator; +import org.apache.flink.api.connector.source.lib.util.IteratorSourceReader; +import org.apache.flink.api.connector.source.lib.util.IteratorSourceSplit; +import org.apache.flink.api.java.typeutils.ResultTypeQueryable; +import org.apache.flink.core.io.SimpleVersionedSerializer; +import org.apache.iceberg.relocated.com.google.common.base.Preconditions; +import org.apache.iceberg.relocated.com.google.common.collect.ImmutableList; +import org.apache.iceberg.relocated.com.google.common.collect.Lists; + +/** + * Implementation of the Source V2 API which uses an iterator to read the elements, and uses a + * single thread to do so. + * + * @param <T> The return type of the source + */ +public abstract class SingleThreadedIteratorSource<T> + implements Source< + T, + SingleThreadedIteratorSource.GlobalSplit<T>, + Collection<SingleThreadedIteratorSource.GlobalSplit<T>>>, + ResultTypeQueryable<T> { + private static final String PARALLELISM_ERROR = "Parallelism should be set to 1"; + + /** + * Creates the iterator to return the elements which then emitted by the source. + * + * @return iterator for the elements + */ + abstract Iterator<T> createIterator(); + + /** + * Serializes the iterator, which is used to save and restore the state of the source. + * + * @return serializer for the iterator + */ + abstract SimpleVersionedSerializer<Iterator<T>> getIteratorSerializer(); + + @Override + public SplitEnumerator<GlobalSplit<T>, Collection<GlobalSplit<T>>> createEnumerator( + SplitEnumeratorContext<GlobalSplit<T>> enumContext) { + Preconditions.checkArgument(enumContext.currentParallelism() == 1, PARALLELISM_ERROR); + return new IteratorSourceEnumerator<>( + enumContext, ImmutableList.of(new GlobalSplit<>(createIterator()))); + } + + @Override + public SplitEnumerator<GlobalSplit<T>, Collection<GlobalSplit<T>>> restoreEnumerator( + SplitEnumeratorContext<GlobalSplit<T>> enumContext, Collection<GlobalSplit<T>> checkpoint) { + Preconditions.checkArgument(enumContext.currentParallelism() == 1, PARALLELISM_ERROR); + return new IteratorSourceEnumerator<>(enumContext, checkpoint); + } + + @Override + public SimpleVersionedSerializer<GlobalSplit<T>> getSplitSerializer() { + return new SplitSerializer<>(getIteratorSerializer()); + } + + @Override + public SimpleVersionedSerializer<Collection<GlobalSplit<T>>> getEnumeratorCheckpointSerializer() { + return new CheckpointSerializer<>(getIteratorSerializer()); + } + + @Override + public SourceReader<T, GlobalSplit<T>> createReader(SourceReaderContext readerContext) + throws Exception { + Preconditions.checkArgument(readerContext.getIndexOfSubtask() == 0, PARALLELISM_ERROR); + return new IteratorSourceReader<>(readerContext); + } + + /** The single split of the {@link SingleThreadedIteratorSource}. */ Review Comment: maybe explain why it is a global split with a constant splitId. if my understanding is correctly, this is related to the per-split watermark. in this case, we only need the global watermark. ########## flink/v1.19/flink/src/main/java/org/apache/iceberg/flink/maintenance/operator/MonitorSource.java: ########## @@ -0,0 +1,200 @@ +/* + * 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.maintenance.operator; + +import java.io.IOException; +import java.util.Iterator; +import org.apache.flink.api.common.typeinfo.TypeInformation; +import org.apache.flink.api.connector.source.Boundedness; +import org.apache.flink.api.connector.source.SourceReader; +import org.apache.flink.api.connector.source.SourceReaderContext; +import org.apache.flink.api.connector.source.util.ratelimit.RateLimitedSourceReader; +import org.apache.flink.api.connector.source.util.ratelimit.RateLimiter; +import org.apache.flink.api.connector.source.util.ratelimit.RateLimiterStrategy; +import org.apache.flink.core.io.SimpleVersionedSerializer; +import org.apache.flink.core.memory.DataInputDeserializer; +import org.apache.flink.core.memory.DataOutputSerializer; +import org.apache.iceberg.Snapshot; +import org.apache.iceberg.Table; +import org.apache.iceberg.flink.TableLoader; +import org.apache.iceberg.relocated.com.google.common.annotations.VisibleForTesting; +import org.apache.iceberg.relocated.com.google.common.base.MoreObjects; +import org.apache.iceberg.relocated.com.google.common.base.Preconditions; +import org.slf4j.Logger; +import org.slf4j.LoggerFactory; + +/** Monitors an Iceberg table for changes */ +public class MonitorSource extends SingleThreadedIteratorSource<TableChange> { + private static final Logger LOG = LoggerFactory.getLogger(MonitorSource.class); + + private final TableLoader tableLoader; + private final RateLimiterStrategy rateLimiterStrategy; + private final long maxReadBack; + + /** + * Creates a {@link org.apache.flink.api.connector.source.Source} which monitors an Iceberg table + * for changes. + * + * @param tableLoader used for accessing the table + * @param rateLimiterStrategy limits the frequency the table is checked + * @param maxReadBack sets the number of snapshots read before stopping change collection + */ + public MonitorSource( + TableLoader tableLoader, RateLimiterStrategy rateLimiterStrategy, long maxReadBack) { + Preconditions.checkNotNull(tableLoader, "Table loader should no be null"); + Preconditions.checkNotNull(rateLimiterStrategy, "Rate limiter strategy should no be null"); + Preconditions.checkArgument(maxReadBack > 0, "Need to read at least 1 snapshot to work"); + + this.tableLoader = tableLoader; + this.rateLimiterStrategy = rateLimiterStrategy; + this.maxReadBack = maxReadBack; + } + + @Override + public Boundedness getBoundedness() { + return Boundedness.CONTINUOUS_UNBOUNDED; + } + + @Override + public TypeInformation<TableChange> getProducedType() { + return TypeInformation.of(TableChange.class); + } + + @Override + Iterator<TableChange> createIterator() { + return new SchedulerEventIterator(tableLoader, null, maxReadBack); + } + + @Override + SimpleVersionedSerializer<Iterator<TableChange>> getIteratorSerializer() { + return new SchedulerEventIteratorSerializer(tableLoader, maxReadBack); + } + + @Override + public SourceReader<TableChange, GlobalSplit<TableChange>> createReader( + SourceReaderContext readerContext) throws Exception { + RateLimiter rateLimiter = rateLimiterStrategy.createRateLimiter(1); + return new RateLimitedSourceReader<>(super.createReader(readerContext), rateLimiter); Review Comment: then we probably don't need the iterator reader. enumerator just emit a global split when appropriate. ########## flink/v1.19/flink/src/main/java/org/apache/iceberg/flink/maintenance/operator/MonitorSource.java: ########## @@ -0,0 +1,200 @@ +/* + * 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.maintenance.operator; + +import java.io.IOException; +import java.util.Iterator; +import org.apache.flink.api.common.typeinfo.TypeInformation; +import org.apache.flink.api.connector.source.Boundedness; +import org.apache.flink.api.connector.source.SourceReader; +import org.apache.flink.api.connector.source.SourceReaderContext; +import org.apache.flink.api.connector.source.util.ratelimit.RateLimitedSourceReader; +import org.apache.flink.api.connector.source.util.ratelimit.RateLimiter; +import org.apache.flink.api.connector.source.util.ratelimit.RateLimiterStrategy; +import org.apache.flink.core.io.SimpleVersionedSerializer; +import org.apache.flink.core.memory.DataInputDeserializer; +import org.apache.flink.core.memory.DataOutputSerializer; +import org.apache.iceberg.Snapshot; +import org.apache.iceberg.Table; +import org.apache.iceberg.flink.TableLoader; +import org.apache.iceberg.relocated.com.google.common.annotations.VisibleForTesting; +import org.apache.iceberg.relocated.com.google.common.base.MoreObjects; +import org.apache.iceberg.relocated.com.google.common.base.Preconditions; +import org.slf4j.Logger; +import org.slf4j.LoggerFactory; + +/** Monitors an Iceberg table for changes */ +public class MonitorSource extends SingleThreadedIteratorSource<TableChange> { + private static final Logger LOG = LoggerFactory.getLogger(MonitorSource.class); + + private final TableLoader tableLoader; + private final RateLimiterStrategy rateLimiterStrategy; + private final long maxReadBack; + + /** + * Creates a {@link org.apache.flink.api.connector.source.Source} which monitors an Iceberg table + * for changes. + * + * @param tableLoader used for accessing the table + * @param rateLimiterStrategy limits the frequency the table is checked + * @param maxReadBack sets the number of snapshots read before stopping change collection + */ + public MonitorSource( + TableLoader tableLoader, RateLimiterStrategy rateLimiterStrategy, long maxReadBack) { + Preconditions.checkNotNull(tableLoader, "Table loader should no be null"); + Preconditions.checkNotNull(rateLimiterStrategy, "Rate limiter strategy should no be null"); + Preconditions.checkArgument(maxReadBack > 0, "Need to read at least 1 snapshot to work"); + + this.tableLoader = tableLoader; + this.rateLimiterStrategy = rateLimiterStrategy; + this.maxReadBack = maxReadBack; + } + + @Override + public Boundedness getBoundedness() { + return Boundedness.CONTINUOUS_UNBOUNDED; + } + + @Override + public TypeInformation<TableChange> getProducedType() { + return TypeInformation.of(TableChange.class); + } + + @Override + Iterator<TableChange> createIterator() { + return new SchedulerEventIterator(tableLoader, null, maxReadBack); + } + + @Override + SimpleVersionedSerializer<Iterator<TableChange>> getIteratorSerializer() { + return new SchedulerEventIteratorSerializer(tableLoader, maxReadBack); + } + + @Override + public SourceReader<TableChange, GlobalSplit<TableChange>> createReader( + SourceReaderContext readerContext) throws Exception { + RateLimiter rateLimiter = rateLimiterStrategy.createRateLimiter(1); + return new RateLimitedSourceReader<>(super.createReader(readerContext), rateLimiter); Review Comment: why don't we just let the enumerator leverage the coordinator single thread pool to schedule the action. Using rate limiter reader seems unnecessary. reader will be idle if there is no split. ``` enumeratorContext.callAsync( this::discoverSplits, this::processDiscoveredSplits, 0L, scanContext.monitorInterval().toMillis()); ``` ########## flink/v1.19/flink/src/main/java/org/apache/iceberg/flink/maintenance/operator/MonitorSource.java: ########## @@ -0,0 +1,200 @@ +/* + * 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.maintenance.operator; + +import java.io.IOException; +import java.util.Iterator; +import org.apache.flink.api.common.typeinfo.TypeInformation; +import org.apache.flink.api.connector.source.Boundedness; +import org.apache.flink.api.connector.source.SourceReader; +import org.apache.flink.api.connector.source.SourceReaderContext; +import org.apache.flink.api.connector.source.util.ratelimit.RateLimitedSourceReader; +import org.apache.flink.api.connector.source.util.ratelimit.RateLimiter; +import org.apache.flink.api.connector.source.util.ratelimit.RateLimiterStrategy; +import org.apache.flink.core.io.SimpleVersionedSerializer; +import org.apache.flink.core.memory.DataInputDeserializer; +import org.apache.flink.core.memory.DataOutputSerializer; +import org.apache.iceberg.Snapshot; +import org.apache.iceberg.Table; +import org.apache.iceberg.flink.TableLoader; +import org.apache.iceberg.relocated.com.google.common.annotations.VisibleForTesting; +import org.apache.iceberg.relocated.com.google.common.base.MoreObjects; +import org.apache.iceberg.relocated.com.google.common.base.Preconditions; +import org.slf4j.Logger; +import org.slf4j.LoggerFactory; + +/** Monitors an Iceberg table for changes */ +public class MonitorSource extends SingleThreadedIteratorSource<TableChange> { + private static final Logger LOG = LoggerFactory.getLogger(MonitorSource.class); + + private final TableLoader tableLoader; + private final RateLimiterStrategy rateLimiterStrategy; + private final long maxReadBack; + + /** + * Creates a {@link org.apache.flink.api.connector.source.Source} which monitors an Iceberg table + * for changes. + * + * @param tableLoader used for accessing the table + * @param rateLimiterStrategy limits the frequency the table is checked + * @param maxReadBack sets the number of snapshots read before stopping change collection + */ + public MonitorSource( + TableLoader tableLoader, RateLimiterStrategy rateLimiterStrategy, long maxReadBack) { + Preconditions.checkNotNull(tableLoader, "Table loader should no be null"); + Preconditions.checkNotNull(rateLimiterStrategy, "Rate limiter strategy should no be null"); + Preconditions.checkArgument(maxReadBack > 0, "Need to read at least 1 snapshot to work"); + + this.tableLoader = tableLoader; + this.rateLimiterStrategy = rateLimiterStrategy; + this.maxReadBack = maxReadBack; + } + + @Override + public Boundedness getBoundedness() { + return Boundedness.CONTINUOUS_UNBOUNDED; + } + + @Override + public TypeInformation<TableChange> getProducedType() { + return TypeInformation.of(TableChange.class); + } + + @Override + Iterator<TableChange> createIterator() { + return new SchedulerEventIterator(tableLoader, null, maxReadBack); + } + + @Override + SimpleVersionedSerializer<Iterator<TableChange>> getIteratorSerializer() { + return new SchedulerEventIteratorSerializer(tableLoader, maxReadBack); + } + + @Override + public SourceReader<TableChange, GlobalSplit<TableChange>> createReader( + SourceReaderContext readerContext) throws Exception { + RateLimiter rateLimiter = rateLimiterStrategy.createRateLimiter(1); + return new RateLimitedSourceReader<>(super.createReader(readerContext), rateLimiter); + } + + /** The Iterator which returns the latest changes on an Iceberg table. */ + @VisibleForTesting + static class SchedulerEventIterator implements Iterator<TableChange> { + private Long lastSnapshotId; + private final long maxReadBack; + private final Table table; + + SchedulerEventIterator(TableLoader tableLoader, Long lastSnapshotId, long maxReadBack) { + this.lastSnapshotId = lastSnapshotId; + this.maxReadBack = maxReadBack; + tableLoader.open(); + this.table = tableLoader.loadTable(); + } + + @Override + public boolean hasNext() { + return true; + } + + @Override + public TableChange next() { + try { + table.refresh(); + Snapshot currentSnapshot = table.currentSnapshot(); + Long current = currentSnapshot != null ? currentSnapshot.snapshotId() : null; + Long checking = current; + TableChange event = new TableChange(0, 0, 0L, 0L, 0); + long readBack = 0; + while (checking != null && !checking.equals(lastSnapshotId) && ++readBack <= maxReadBack) { + Snapshot snapshot = table.snapshot(checking); + if (snapshot != null) { + LOG.debug("Reading snapshot {}", snapshot.snapshotId()); + event.merge(new TableChange(snapshot, table.io())); + checking = snapshot.parentId(); + } else { + // If the last snapshot has been removed from the history + checking = null; + } + } + + lastSnapshotId = current; + return event; + } catch (Exception e) { + LOG.warn("Failed to fetch table changes for {}", table, e); + return new TableChange(0, 0, 0L, 0L, 0); + } + } + + @Override + public String toString() { + return MoreObjects.toStringHelper(this) + .add("lastSnapshotId", lastSnapshotId) + .add("maxReadBack", maxReadBack) + .add("table", table) + .toString(); + } + } + + private static final class SchedulerEventIteratorSerializer Review Comment: if we change the enumerator threading model, this can become just a enumerator state serializer on last snapshot. ########## flink/v1.19/flink/src/main/java/org/apache/iceberg/flink/maintenance/operator/MonitorSource.java: ########## @@ -0,0 +1,200 @@ +/* + * 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.maintenance.operator; + +import java.io.IOException; +import java.util.Iterator; +import org.apache.flink.api.common.typeinfo.TypeInformation; +import org.apache.flink.api.connector.source.Boundedness; +import org.apache.flink.api.connector.source.SourceReader; +import org.apache.flink.api.connector.source.SourceReaderContext; +import org.apache.flink.api.connector.source.util.ratelimit.RateLimitedSourceReader; +import org.apache.flink.api.connector.source.util.ratelimit.RateLimiter; +import org.apache.flink.api.connector.source.util.ratelimit.RateLimiterStrategy; +import org.apache.flink.core.io.SimpleVersionedSerializer; +import org.apache.flink.core.memory.DataInputDeserializer; +import org.apache.flink.core.memory.DataOutputSerializer; +import org.apache.iceberg.Snapshot; +import org.apache.iceberg.Table; +import org.apache.iceberg.flink.TableLoader; +import org.apache.iceberg.relocated.com.google.common.annotations.VisibleForTesting; +import org.apache.iceberg.relocated.com.google.common.base.MoreObjects; +import org.apache.iceberg.relocated.com.google.common.base.Preconditions; +import org.slf4j.Logger; +import org.slf4j.LoggerFactory; + +/** Monitors an Iceberg table for changes */ +public class MonitorSource extends SingleThreadedIteratorSource<TableChange> { + private static final Logger LOG = LoggerFactory.getLogger(MonitorSource.class); + + private final TableLoader tableLoader; + private final RateLimiterStrategy rateLimiterStrategy; + private final long maxReadBack; + + /** + * Creates a {@link org.apache.flink.api.connector.source.Source} which monitors an Iceberg table + * for changes. + * + * @param tableLoader used for accessing the table + * @param rateLimiterStrategy limits the frequency the table is checked + * @param maxReadBack sets the number of snapshots read before stopping change collection + */ + public MonitorSource( + TableLoader tableLoader, RateLimiterStrategy rateLimiterStrategy, long maxReadBack) { + Preconditions.checkNotNull(tableLoader, "Table loader should no be null"); + Preconditions.checkNotNull(rateLimiterStrategy, "Rate limiter strategy should no be null"); + Preconditions.checkArgument(maxReadBack > 0, "Need to read at least 1 snapshot to work"); + + this.tableLoader = tableLoader; + this.rateLimiterStrategy = rateLimiterStrategy; + this.maxReadBack = maxReadBack; + } + + @Override + public Boundedness getBoundedness() { + return Boundedness.CONTINUOUS_UNBOUNDED; + } + + @Override + public TypeInformation<TableChange> getProducedType() { + return TypeInformation.of(TableChange.class); + } + + @Override + Iterator<TableChange> createIterator() { + return new SchedulerEventIterator(tableLoader, null, maxReadBack); + } + + @Override + SimpleVersionedSerializer<Iterator<TableChange>> getIteratorSerializer() { + return new SchedulerEventIteratorSerializer(tableLoader, maxReadBack); + } + + @Override + public SourceReader<TableChange, GlobalSplit<TableChange>> createReader( + SourceReaderContext readerContext) throws Exception { + RateLimiter rateLimiter = rateLimiterStrategy.createRateLimiter(1); + return new RateLimitedSourceReader<>(super.createReader(readerContext), rateLimiter); + } + + /** The Iterator which returns the latest changes on an Iceberg table. */ + @VisibleForTesting + static class SchedulerEventIterator implements Iterator<TableChange> { Review Comment: nit: it seems that `TableChangeIterator` is a better match -- 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: issues-unsubscr...@iceberg.apache.org For queries about this service, please contact Infrastructure at: us...@infra.apache.org --------------------------------------------------------------------- To unsubscribe, e-mail: issues-unsubscr...@iceberg.apache.org For additional commands, e-mail: issues-h...@iceberg.apache.org
