mxm commented on code in PR #15996: URL: https://github.com/apache/iceberg/pull/15996#discussion_r3225164834
########## flink/v2.1/flink/src/main/java/org/apache/iceberg/flink/maintenance/operator/EqualityConvertPlanner.java: ########## @@ -0,0 +1,595 @@ +/* + * 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.io.UncheckedIOException; +import java.util.Collections; +import java.util.List; +import java.util.Map; +import java.util.Set; +import org.apache.flink.annotation.Internal; +import org.apache.flink.api.common.state.ListState; +import org.apache.flink.api.common.state.ListStateDescriptor; +import org.apache.flink.api.common.typeinfo.Types; +import org.apache.flink.metrics.Counter; +import org.apache.flink.metrics.MetricGroup; +import org.apache.flink.runtime.state.StateInitializationContext; +import org.apache.flink.runtime.state.StateSnapshotContext; +import org.apache.flink.streaming.api.operators.AbstractStreamOperator; +import org.apache.flink.streaming.api.operators.OneInputStreamOperator; +import org.apache.flink.streaming.api.watermark.Watermark; +import org.apache.flink.streaming.runtime.streamrecord.StreamRecord; +import org.apache.flink.util.OutputTag; +import org.apache.iceberg.DataFile; +import org.apache.iceberg.DeleteFile; +import org.apache.iceberg.FileContent; +import org.apache.iceberg.FileScanTask; +import org.apache.iceberg.PartitionSpec; +import org.apache.iceberg.Snapshot; +import org.apache.iceberg.SnapshotChanges; +import org.apache.iceberg.Table; +import org.apache.iceberg.flink.TableLoader; +import org.apache.iceberg.flink.maintenance.api.Trigger; +import org.apache.iceberg.io.CloseableIterable; +import org.apache.iceberg.relocated.com.google.common.base.Preconditions; +import org.apache.iceberg.relocated.com.google.common.collect.Lists; +import org.apache.iceberg.relocated.com.google.common.collect.Maps; +import org.apache.iceberg.relocated.com.google.common.collect.Sets; +import org.apache.iceberg.util.ContentFileUtil; +import org.slf4j.Logger; +import org.slf4j.LoggerFactory; + +/** + * Planner for the equality delete conversion pipeline. For each trigger, it picks the oldest + * staging snapshot that hasn't been converted yet and emits {@link ReadCommand}s describing the + * files its downstream readers and workers must process. + * + * <p>Emissions are grouped into phases separated by watermarks at the boundaries that gate the + * worker's keyed state: + * + * <ol> + * <li>Main data — read once per new equality-field-set to create the worker's key index. + * <li>Equality + positional deletes — eq deletes are resolved against the index in the worker; + * pos deletes bypass the worker via a side output and are part of the same DV merge. + * <li>New staging data — added to the index for the next cycle. + * </ol> + * + * The planner also detects external commits on main (not produced by the converter itself) and + * re-emits main data so the worker can rebuild its index before resolving the current cycle's + * equality deletes. + * + * <p>An {@link EqualityConvertPlanResult} with the current cycle's metadata is emitted via the + * {@link #METADATA_STREAM} side output after the read commands. + */ +@Internal +public class EqualityConvertPlanner extends AbstractStreamOperator<ReadCommand> + implements OneInputStreamOperator<Trigger, ReadCommand> { + + private static final Logger LOG = LoggerFactory.getLogger(EqualityConvertPlanner.class); + + public static final OutputTag<EqualityConvertPlanResult> METADATA_STREAM = + new OutputTag<>("metadata-stream") {}; + + private static final String PROCESSED_EQ_DELETE_FILE_NUM_METRIC = "processedEqDeleteFileNum"; + private static final String PROCESSED_STAGING_SNAPSHOT_NUM_METRIC = "processedStagingSnapshotNum"; + private static final String SKIPPED_NO_OP_CYCLES_METRIC = "skippedNoOpCycles"; + private static final String REINDEX_COUNT_METRIC = "reindexCount"; + + private final String tableName; + private final String taskName; + private final int taskIndex; + private final TableLoader tableLoader; + private final String stagingBranch; + private final String targetBranch; + + // Main snapshot id the worker's index reflects. + private transient ListState<Long> indexSnapshotState; + // Equality field sets for which main data has been emitted. + private transient ListState<List<Integer>> mainIndexEmittedState; + + private transient Table table; + + private transient Long lastMainSnapshotId; + private transient Long lastStagingSnapshotId; + private transient Long indexSnapshotId; + + // Sorted equality-field-id lists for which main data has already been emitted to create the + // worker index. One entry per distinct key schema. Checkpointed so first-time emission survives + // restarts. + private transient Set<List<Integer>> mainIndexEmittedSet; + + private transient long nextPhaseTs; + + private transient Counter processedEqDeleteFileNumCounter; + private transient Counter processedStagingSnapshotNumCounter; + private transient Counter skippedNoOpCyclesCounter; + private transient Counter reindexCounter; + + public EqualityConvertPlanner( + String tableName, + String taskName, + int taskIndex, + TableLoader tableLoader, + String stagingBranch, + String targetBranch) { + this.tableName = tableName; + this.taskName = taskName; + this.taskIndex = taskIndex; + this.tableLoader = tableLoader; + this.stagingBranch = stagingBranch; + this.targetBranch = targetBranch; + } + + @Override + public void open() throws Exception { + super.open(); + if (!tableLoader.isOpen()) { + tableLoader.open(); + } + + table = tableLoader.loadTable(); Review Comment: We now require to set equality columns on the builder level and use those for creating the index, even if there are no staging commits. -- 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]
