[GitHub] [hudi] danny0405 commented on a change in pull request #4679: [HUDI-3315] RFC-35 Part-1 Support bucket index in Flink writer

[GitBox <gi...@apache.org>]

danny0405 commented on a change in pull request #4679:
URL: https://github.com/apache/hudi/pull/4679#discussion_r811607678



##########
File path: hudi-flink/src/main/java/org/apache/hudi/configuration/FlinkOptions.java
##########
@@ -106,6 +108,12 @@ private FlinkOptions() {
   // ------------------------------------------------------------------------
   //  Index Options
   // ------------------------------------------------------------------------
+  public static final ConfigOption<String> INDEX_TYPE = ConfigOptions
+      .key("write.index.type")
+      .stringType()

Review comment:
       `write.index.type` => `index.type` ?

##########
File path: hudi-flink/src/main/java/org/apache/hudi/configuration/FlinkOptions.java
##########
@@ -310,6 +318,20 @@ private FlinkOptions() {
           + "Actual value will be obtained by invoking .toString() on the field value. Nested fields can be specified using "
           + "the dot notation eg: `a.b.c`");
 
+  public static final ConfigOption<String> INDEX_KEY_FIELD = ConfigOptions
+      .key(HoodieIndexConfig.BUCKET_INDEX_HASH_FIELD.key())
+      .stringType()
+      .defaultValue("uuid")
+      .withDescription("Record key field. Value to be used as the `recordKey` component of `HoodieKey`.\n"
+        + "Actual value will be obtained by invoking .toString() on the field value. Nested fields can be specified using "
+        + "the dot notation eg: `a.b.c`");
+
+  public static final ConfigOption<Integer> BUCKET_INDEX_NUM_BUCKETS = ConfigOptions
+      .key(HoodieIndexConfig.BUCKET_INDEX_NUM_BUCKETS.key())
+      .intType()
+      .defaultValue(256) // default 256 buckets per partition
+      .withDescription("Hudi bucket number per partition. Only affected if using Hudi bucket index.");

Review comment:
       If people change the `BUCKET_INDEX_NUM_BUCKETS` or the write function parallelism, does the hash index still work ?

##########
File path: hudi-flink/src/main/java/org/apache/hudi/sink/BucketStreamWriteFunction.java
##########
@@ -0,0 +1,158 @@
+/*
+ * 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.hudi.sink;
+
+import org.apache.flink.configuration.Configuration;
+import org.apache.flink.runtime.state.FunctionInitializationContext;
+import org.apache.flink.streaming.api.functions.ProcessFunction;
+import org.apache.flink.util.Collector;
+import org.apache.hudi.common.model.FileSlice;
+import org.apache.hudi.common.model.HoodieKey;
+import org.apache.hudi.common.model.HoodieRecord;
+import org.apache.hudi.common.model.HoodieRecordLocation;
+import org.apache.hudi.common.table.timeline.HoodieInstant;
+import org.apache.hudi.common.util.Option;
+import org.apache.hudi.configuration.FlinkOptions;
+import org.apache.hudi.index.bucket.BucketIdentifier;
+import org.apache.hudi.table.HoodieFlinkTable;
+import org.slf4j.Logger;
+import org.slf4j.LoggerFactory;
+
+import java.io.IOException;
+import java.util.HashMap;
+import java.util.HashSet;
+import java.util.List;
+
+import static java.util.stream.Collectors.toList;
+
+public class BucketStreamWriteFunction<I> extends StreamWriteFunction<I> {
+
+  private static final Logger LOG = LoggerFactory.getLogger(BucketStreamWriteFunction.class);
+
+  private int taskTotalParallelism;
+
+  private int currentParallelism;
+
+  private int hiveBucketNum;
+
+  protected transient HoodieFlinkTable table;
+
+  private String indexKeyFields;
+
+  private HashMap<String, String> bucketToFileIDMap;
+
+  /**
+   * Constructs a BucketStreamWriteFunction.
+   *
+   * @param config The config options
+   */
+  public BucketStreamWriteFunction(Configuration config) {
+    super(config);
+    this.bucketToFileIDMap = new HashMap<String, String>();
+  }
+
+  @Override
+  public void open(Configuration parameters) throws IOException {
+    super.open(parameters);
+    this.hiveBucketNum = config.getInteger(FlinkOptions.BUCKET_INDEX_NUM_BUCKETS);
+    this.indexKeyFields = config.getString(FlinkOptions.INDEX_KEY_FIELD);
+    this.taskID = getRuntimeContext().getIndexOfThisSubtask();
+    this.currentParallelism = getRuntimeContext().getNumberOfParallelSubtasks();
+    this.taskTotalParallelism = getRuntimeContext().getMaxNumberOfParallelSubtasks();
+    bootstrapIndex();
+  }
+
+  @Override
+  public void initializeState(FunctionInitializationContext context) throws Exception {
+    super.initializeState(context);
+    this.table = this.writeClient.getHoodieTable();
+  }
+
+  @Override
+  public void processElement(I i, ProcessFunction<I, Object>.Context context, Collector<Object> collector) throws Exception {
+    HoodieRecord<?> record = (HoodieRecord<?>) i;
+    final HoodieKey hoodieKey = record.getKey();
+    final HoodieRecordLocation location;
+
+    final int bucketNum = BucketIdentifier.getBucketId(hoodieKey, indexKeyFields, hiveBucketNum);
+    final String partitionBucketId = BucketIdentifier.partitionBucketIdStr(hoodieKey.getPartitionPath(), bucketNum);
+
+    if (bucketToFileIDMap.containsKey(partitionBucketId)) {
+      location = new HoodieRecordLocation("U", bucketToFileIDMap.get(partitionBucketId));
+    } else {
+      String newFileId = BucketIdentifier.newBucketFileIdPrefix(bucketNum);
+      location = new HoodieRecordLocation("I", newFileId);
+      bucketToFileIDMap.put(partitionBucketId, newFileId);
+    }
+    record.unseal();
+    record.setCurrentLocation(location);
+    record.seal();
+    bufferRecord(record);
+  }
+
+  /**
+   * Get partition_bucket -> fileID mapping from the existing hudi table.
+   * This is a required operation for each restart to avoid having duplicate file ids for one bucket.
+   */
+  private void bootstrapIndex() throws IOException {
+    Option<HoodieInstant> latestCommitTime = table.getFileSystemView().getTimeline().filterCompletedInstants().lastInstant();
+    if (!latestCommitTime.isPresent()) {
+      return;
+    }
+    // bootstrap bucket info from existing file system
+    // bucketNum % totalParallelism == this taskID belongs to this task
+    HashSet<Integer> bucketToLoad = new HashSet<>();
+    for (int i = 0; i < hiveBucketNum; i++) {
+      int partitionOfBucket = BucketIdentifier.mod(i, currentParallelism);
+      if (partitionOfBucket == taskID) {
+        LOG.info(String.format("Bootstrapping index. Adding bucket %s , "
+            + "Current parallelism: %s , Max parallelism: %s , Current task id: %s",
+            i, currentParallelism, taskTotalParallelism, taskID));
+        bucketToLoad.add(i);
+      }
+    }
+    bucketToLoad.stream().forEach(bucket -> LOG.info(String.format("bucketToLoad contains %s", bucket)));
+
+    LOG.info(String.format("Loading Hoodie Table %s, with path %s", table.getMetaClient().getTableConfig().getTableName(),
+        table.getMetaClient().getBasePath()));
+
+    // Iterate through all existing partitions to load existing fileID belongs to this task
+    List<String> partitions = table.getMetadata().getAllPartitionPaths();
+    for (String partitionPath : partitions) {
+      List<FileSlice> latestFileSlices = table.getSliceView()
+          .getLatestFileSlices(partitionPath)
+          .collect(toList());
+      for (FileSlice fileslice : latestFileSlices) {
+        String fileID = fileslice.getFileId();
+        int bucketNumber = BucketIdentifier.bucketIdFromFileId(fileID);
+        if (bucketToLoad.contains(bucketNumber)) {
+          String partitionBucketId = BucketIdentifier.partitionBucketIdStr(partitionPath, bucketNumber);
+          LOG.info(String.format("Should load this partition bucket %s with fileID %s", partitionBucketId, fileID));
+          if (bucketToFileIDMap.containsKey(partitionBucketId)) {
+            throw new RuntimeException(String.format("Duplicate fileID %s from partitionBucket %s found "
+              + "during the fileGroupPerPartitionedBucketState initialization.", fileID, partitionBucketId));
+          } else {
+            LOG.info(String.format("Adding fileID %s to the partition bucket %s.", fileID, partitionBucketId));
+            bucketToFileIDMap.put(partitionBucketId, fileID);
+          }

Review comment:
       The bucketToFileIDMap seems never be cleared, is there possibility that this map be put into the state ?

##########
File path: hudi-flink/src/main/java/org/apache/hudi/configuration/OptionsResolver.java
##########
@@ -101,6 +102,10 @@ public static boolean isPartitionedTable(Configuration conf) {
     return FilePathUtils.extractPartitionKeys(conf).length > 0;
   }
 
+  public static boolean isBucketIndexTable(Configuration conf) {
+    return conf.getString(FlinkOptions.INDEX_TYPE).equals(HoodieIndex.IndexType.BUCKET.name());

Review comment:
       `isBucketIndexTable` => `isBucketIndexType`

##########
File path: hudi-flink/src/main/java/org/apache/hudi/sink/BucketStreamWriteFunction.java
##########
@@ -0,0 +1,158 @@
+/*
+ * 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.hudi.sink;
+
+import org.apache.flink.configuration.Configuration;
+import org.apache.flink.runtime.state.FunctionInitializationContext;
+import org.apache.flink.streaming.api.functions.ProcessFunction;
+import org.apache.flink.util.Collector;
+import org.apache.hudi.common.model.FileSlice;
+import org.apache.hudi.common.model.HoodieKey;
+import org.apache.hudi.common.model.HoodieRecord;
+import org.apache.hudi.common.model.HoodieRecordLocation;
+import org.apache.hudi.common.table.timeline.HoodieInstant;
+import org.apache.hudi.common.util.Option;
+import org.apache.hudi.configuration.FlinkOptions;
+import org.apache.hudi.index.bucket.BucketIdentifier;
+import org.apache.hudi.table.HoodieFlinkTable;
+import org.slf4j.Logger;
+import org.slf4j.LoggerFactory;
+
+import java.io.IOException;
+import java.util.HashMap;
+import java.util.HashSet;
+import java.util.List;
+
+import static java.util.stream.Collectors.toList;
+
+public class BucketStreamWriteFunction<I> extends StreamWriteFunction<I> {
+
+  private static final Logger LOG = LoggerFactory.getLogger(BucketStreamWriteFunction.class);
+
+  private int taskTotalParallelism;
+
+  private int currentParallelism;
+
+  private int hiveBucketNum;
+
+  protected transient HoodieFlinkTable table;
+
+  private String indexKeyFields;
+
+  private HashMap<String, String> bucketToFileIDMap;
+
+  /**
+   * Constructs a BucketStreamWriteFunction.
+   *
+   * @param config The config options
+   */
+  public BucketStreamWriteFunction(Configuration config) {
+    super(config);
+    this.bucketToFileIDMap = new HashMap<String, String>();
+  }
+
+  @Override
+  public void open(Configuration parameters) throws IOException {
+    super.open(parameters);
+    this.hiveBucketNum = config.getInteger(FlinkOptions.BUCKET_INDEX_NUM_BUCKETS);
+    this.indexKeyFields = config.getString(FlinkOptions.INDEX_KEY_FIELD);
+    this.taskID = getRuntimeContext().getIndexOfThisSubtask();
+    this.currentParallelism = getRuntimeContext().getNumberOfParallelSubtasks();
+    this.taskTotalParallelism = getRuntimeContext().getMaxNumberOfParallelSubtasks();
+    bootstrapIndex();

Review comment:
       `this.currentParallelism` => `this.parallelism`
   `this.taskTotalParallelism` => `this.maxParallelism`

##########
File path: hudi-flink/src/main/java/org/apache/hudi/configuration/FlinkOptions.java
##########
@@ -310,6 +318,20 @@ private FlinkOptions() {
           + "Actual value will be obtained by invoking .toString() on the field value. Nested fields can be specified using "
           + "the dot notation eg: `a.b.c`");
 
+  public static final ConfigOption<String> INDEX_KEY_FIELD = ConfigOptions
+      .key(HoodieIndexConfig.BUCKET_INDEX_HASH_FIELD.key())
+      .stringType()

Review comment:
       Does this key must be same same with the primary key ? Because all the changes of a key must belong to one data bucket.

##########
File path: hudi-flink/src/main/java/org/apache/hudi/sink/common/AbstractStreamWriteFunction.java
##########
@@ -107,7 +107,7 @@
   /**
    * List state of the write metadata events.
    */
-  private transient ListState<WriteMetadataEvent> writeMetadataState;
+  protected transient ListState<WriteMetadataEvent> writeMetadataState;
 

Review comment:
       Seems not used by sub-class ?




-- 
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: commits-unsubscribe@hudi.apache.org

For queries about this service, please contact Infrastructure at:
users@infra.apache.org