You are viewing a plain text version of this content. The canonical link for it is here.
Posted to commits@hudi.apache.org by GitBox <gi...@apache.org> on 2021/01/06 14:19:18 UTC
[GitHub] [hudi] garyli1019 commented on a change in pull request #2375: [HUDI-1332] Introduce FlinkHoodieBloomIndex to hudi-flink-client
garyli1019 commented on a change in pull request #2375:
URL: https://github.com/apache/hudi/pull/2375#discussion_r552648364
##########
File path: hudi-client/hudi-flink-client/src/test/java/org/apache/hudi/index/bloom/TestFlinkHoodieBloomIndex.java
##########
@@ -0,0 +1,444 @@
+/*
+ * 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.index.bloom;
+
+import org.apache.hudi.common.model.HoodieKey;
+import org.apache.hudi.common.model.HoodieRecord;
+import org.apache.hudi.common.testutils.RawTripTestPayload;
+import org.apache.hudi.common.util.Option;
+import org.apache.hudi.common.util.collection.Pair;
+import org.apache.hudi.table.HoodieFlinkTable;
+import org.apache.hudi.table.HoodieTable;
+import scala.Tuple2;
+
+import java.util.ArrayList;
+import java.util.HashMap;
+import java.util.List;
+import java.util.Map;
+
+import static java.util.Arrays.asList;
+import static org.apache.hudi.common.testutils.SchemaTestUtil.getSchemaFromResource;
+import static org.junit.jupiter.api.Assertions.*;
+
+public class TestFlinkHoodieBloomIndex extends org.apache.hudi.testutils.FlinkHoodieClientTestHarness {
Review comment:
can we move all of `org.apache.hudi.testutils.FlinkHoodieClientTestHarness` style to the import?
##########
File path: hudi-client/hudi-flink-client/src/test/java/org/apache/hudi/testutils/FlinkHoodieClientTestHarness.java
##########
@@ -0,0 +1,239 @@
+/*
+ * 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.testutils;
+
+import org.apache.hudi.client.FlinkTaskContextSupplier;
+import org.apache.hudi.client.HoodieFlinkWriteClient;
+import org.apache.hudi.common.table.HoodieTableMetaClient;
+import org.apache.hudi.common.testutils.HoodieCommonTestHarness;
+
+import java.io.IOException;
+import java.util.List;
+import java.util.Map;
+
+/**
+ * The test harness for resource initialization and cleanup.
+ */
+public abstract class FlinkHoodieClientTestHarness extends HoodieCommonTestHarness implements java.io.Serializable {
Review comment:
Can we use the existing `HoodieFlinkClientTestHarness`. This was merged recently
##########
File path: hudi-client/hudi-flink-client/src/test/java/org/apache/hudi/index/bloom/TestFlinkHoodieBloomIndex.java
##########
@@ -0,0 +1,444 @@
+/*
+ * 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.index.bloom;
+
+import org.apache.hudi.common.model.HoodieKey;
+import org.apache.hudi.common.model.HoodieRecord;
+import org.apache.hudi.common.testutils.RawTripTestPayload;
+import org.apache.hudi.common.util.Option;
+import org.apache.hudi.common.util.collection.Pair;
+import org.apache.hudi.table.HoodieFlinkTable;
+import org.apache.hudi.table.HoodieTable;
+import scala.Tuple2;
+
+import java.util.ArrayList;
+import java.util.HashMap;
+import java.util.List;
+import java.util.Map;
+
+import static java.util.Arrays.asList;
+import static org.apache.hudi.common.testutils.SchemaTestUtil.getSchemaFromResource;
+import static org.junit.jupiter.api.Assertions.*;
+
+public class TestFlinkHoodieBloomIndex extends org.apache.hudi.testutils.FlinkHoodieClientTestHarness {
+
+ private static final org.apache.avro.Schema SCHEMA = getSchemaFromResource(TestFlinkHoodieBloomIndex.class, "/exampleSchema.avsc", true);
+ private static final String TEST_NAME_WITH_PARAMS = "[{index}] Test with rangePruning={0}, treeFiltering={1}, bucketizedChecking={2}";
+
+ public static java.util.stream.Stream<org.junit.jupiter.params.provider.Arguments> configParams() {
+ Object[][] data =
+ new Object[][] {{true, true, true}, {false, true, true}, {true, true, false}, {true, false, true}};
+ return java.util.stream.Stream.of(data).map(org.junit.jupiter.params.provider.Arguments::of);
+ }
+
+ @org.junit.jupiter.api.BeforeEach
+ public void setUp() throws Exception {
+ initFlinkContexts();
+ initPath();
+ initFileSystem();
+ // We have some records to be tagged (two different partitions)
+ initMetaClient();
+ }
+
+ @org.junit.jupiter.api.AfterEach
+ public void tearDown() throws Exception {
+ cleanupResources();
+ }
+
+ private org.apache.hudi.config.HoodieWriteConfig makeConfig(boolean rangePruning, boolean treeFiltering, boolean bucketizedChecking) {
+ return org.apache.hudi.config.HoodieWriteConfig.newBuilder().withPath(basePath)
+ .withIndexConfig(org.apache.hudi.config.HoodieIndexConfig.newBuilder().bloomIndexPruneByRanges(rangePruning)
+ .bloomIndexTreebasedFilter(treeFiltering).bloomIndexBucketizedChecking(bucketizedChecking)
+ .bloomIndexKeysPerBucket(2).build())
+ .build();
+ }
+
+ @org.junit.jupiter.params.ParameterizedTest(name = TEST_NAME_WITH_PARAMS)
+ @org.junit.jupiter.params.provider.MethodSource("configParams")
+ public void testLoadInvolvedFiles(boolean rangePruning, boolean treeFiltering, boolean bucketizedChecking) throws Exception {
+ org.apache.hudi.config.HoodieWriteConfig config = makeConfig(rangePruning, treeFiltering, bucketizedChecking);
+ FlinkHoodieBloomIndex index = new FlinkHoodieBloomIndex(config);
+ HoodieTable hoodieTable = HoodieFlinkTable.create(config, context, metaClient);
+ org.apache.hudi.testutils.HoodieFlinkWriteableTestTable testTable = org.apache.hudi.testutils.HoodieFlinkWriteableTestTable.of(hoodieTable, SCHEMA);
+
+ // Create some partitions, and put some files
+ // "2016/01/21": 0 file
+ // "2016/04/01": 1 file (2_0_20160401010101.parquet)
+ // "2015/03/12": 3 files (1_0_20150312101010.parquet, 3_0_20150312101010.parquet, 4_0_20150312101010.parquet)
+ testTable.withPartitionMetaFiles("2016/01/21", "2016/04/01", "2015/03/12");
+
+ RawTripTestPayload rowChange1 =
+ new RawTripTestPayload("{\"_row_key\":\"000\",\"time\":\"2016-01-31T03:16:41.415Z\",\"number\":12}");
+ HoodieRecord record1 =
+ new HoodieRecord(new HoodieKey(rowChange1.getRowKey(), rowChange1.getPartitionPath()), rowChange1);
+ RawTripTestPayload rowChange2 =
+ new RawTripTestPayload("{\"_row_key\":\"001\",\"time\":\"2016-01-31T03:16:41.415Z\",\"number\":12}");
+ HoodieRecord record2 =
+ new HoodieRecord(new HoodieKey(rowChange2.getRowKey(), rowChange2.getPartitionPath()), rowChange2);
+ RawTripTestPayload rowChange3 =
+ new RawTripTestPayload("{\"_row_key\":\"002\",\"time\":\"2016-01-31T03:16:41.415Z\",\"number\":12}");
+ HoodieRecord record3 =
+ new HoodieRecord(new HoodieKey(rowChange3.getRowKey(), rowChange3.getPartitionPath()), rowChange3);
+ RawTripTestPayload rowChange4 =
+ new RawTripTestPayload("{\"_row_key\":\"003\",\"time\":\"2016-01-31T03:16:41.415Z\",\"number\":12}");
+ HoodieRecord record4 =
+ new HoodieRecord(new HoodieKey(rowChange4.getRowKey(), rowChange4.getPartitionPath()), rowChange4);
+
+ java.util.List<String> partitions = asList("2016/01/21", "2016/04/01", "2015/03/12");
+ java.util.List<scala.Tuple2<String, BloomIndexFileInfo>> filesList = index.loadInvolvedFiles(partitions, context, hoodieTable);
+ // Still 0, as no valid commit
+ assertEquals(0, filesList.size());
+
+ testTable.addCommit("20160401010101").withInserts("2016/04/01", "2");
+ testTable.addCommit("20150312101010").withInserts("2015/03/12", "1")
+ .withInserts("2015/03/12", "3", record1)
+ .withInserts("2015/03/12", "4", record2, record3, record4);
+ metaClient.reloadActiveTimeline();
+
+ filesList = index.loadInvolvedFiles(partitions, context, hoodieTable);
+ assertEquals(4, filesList.size());
+
+ if (rangePruning) {
+ // these files will not have the key ranges
+ assertNull(filesList.get(0)._2().getMaxRecordKey());
+ assertNull(filesList.get(0)._2().getMinRecordKey());
+ assertFalse(filesList.get(1)._2().hasKeyRanges());
+ assertNotNull(filesList.get(2)._2().getMaxRecordKey());
+ assertNotNull(filesList.get(2)._2().getMinRecordKey());
+ assertTrue(filesList.get(3)._2().hasKeyRanges());
+
+ // no longer sorted, but should have same files.
+
+ java.util.List<scala.Tuple2<String, BloomIndexFileInfo>> expected =
+ asList(new Tuple2<>("2016/04/01", new BloomIndexFileInfo("2")),
+ new Tuple2<>("2015/03/12", new BloomIndexFileInfo("1")),
+ new Tuple2<>("2015/03/12", new BloomIndexFileInfo("3", "000", "000")),
+ new Tuple2<>("2015/03/12", new BloomIndexFileInfo("4", "001", "003")));
+ assertEquals(expected, filesList);
+ }
+ }
+
+ @org.junit.jupiter.params.ParameterizedTest(name = TEST_NAME_WITH_PARAMS)
+ @org.junit.jupiter.params.provider.MethodSource("configParams")
+ public void testRangePruning(boolean rangePruning, boolean treeFiltering, boolean bucketizedChecking) {
+ org.apache.hudi.config.HoodieWriteConfig config = makeConfig(rangePruning, treeFiltering, bucketizedChecking);
+ FlinkHoodieBloomIndex index = new FlinkHoodieBloomIndex(config);
+
+ final java.util.Map<String, java.util.List<BloomIndexFileInfo>> partitionToFileIndexInfo = new HashMap<>();
+ partitionToFileIndexInfo.put("2017/10/22",
+ asList(new BloomIndexFileInfo("f1"), new BloomIndexFileInfo("f2", "000", "000"),
+ new BloomIndexFileInfo("f3", "001", "003"), new BloomIndexFileInfo("f4", "002", "007"),
+ new BloomIndexFileInfo("f5", "009", "010")));
+
+ Map<String, List<String>> partitionRecordKeyMap = new HashMap<>();
+ asList(new Tuple2<>("2017/10/22", "003"), new Tuple2<>("2017/10/22", "002"),
+ new Tuple2<>("2017/10/22", "005"), new Tuple2<>("2017/10/22", "004"))
+ .forEach( t -> {
+ List<String> recordKeyList = partitionRecordKeyMap.getOrDefault(t._1, new ArrayList<>());
+ recordKeyList.add(t._2);
+ partitionRecordKeyMap.put(t._1, recordKeyList);
+ });
+
+ List<scala.Tuple2<String, HoodieKey>> comparisonKeyList =
+ index.explodeRecordsWithFileComparisons(partitionToFileIndexInfo, partitionRecordKeyMap);
+
+ assertEquals(10, comparisonKeyList.size());
+ java.util.Map<String, java.util.List<String>> recordKeyToFileComps = comparisonKeyList.stream()
+ .collect(java.util.stream.Collectors.groupingBy(t -> t._2.getRecordKey(), java.util.stream.Collectors.mapping(t -> t._1, java.util.stream.Collectors.toList())));
+
+ assertEquals(4, recordKeyToFileComps.size());
+ assertEquals(new java.util.HashSet<>(asList("f1", "f3", "f4")), new java.util.HashSet<>(recordKeyToFileComps.get("002")));
+ assertEquals(new java.util.HashSet<>(asList("f1", "f3", "f4")), new java.util.HashSet<>(recordKeyToFileComps.get("003")));
+ assertEquals(new java.util.HashSet<>(asList("f1", "f4")), new java.util.HashSet<>(recordKeyToFileComps.get("004")));
+ assertEquals(new java.util.HashSet<>(asList("f1", "f4")), new java.util.HashSet<>(recordKeyToFileComps.get("005")));
+ }
+
+ @org.junit.jupiter.api.Test
+ public void testCheckUUIDsAgainstOneFile() throws Exception {
+ final String partition = "2016/01/31";
+ // Create some records to use
+ String recordStr1 = "{\"_row_key\":\"1eb5b87a-1feh-4edd-87b4-6ec96dc405a0\","
+ + "\"time\":\"2016-01-31T03:16:41.415Z\",\"number\":12}";
+ String recordStr2 = "{\"_row_key\":\"2eb5b87b-1feu-4edd-87b4-6ec96dc405a0\","
+ + "\"time\":\"2016-01-31T03:20:41.415Z\",\"number\":100}";
+ String recordStr3 = "{\"_row_key\":\"3eb5b87c-1fej-4edd-87b4-6ec96dc405a0\","
+ + "\"time\":\"2016-01-31T03:16:41.415Z\",\"number\":15}";
+ String recordStr4 = "{\"_row_key\":\"4eb5b87c-1fej-4edd-87b4-6ec96dc405a0\","
+ + "\"time\":\"2016-01-31T03:16:41.415Z\",\"number\":32}";
+ RawTripTestPayload rowChange1 = new RawTripTestPayload(recordStr1);
+ HoodieRecord record1 =
+ new HoodieRecord(new HoodieKey(rowChange1.getRowKey(), rowChange1.getPartitionPath()), rowChange1);
+ RawTripTestPayload rowChange2 = new RawTripTestPayload(recordStr2);
+ HoodieRecord record2 =
+ new HoodieRecord(new HoodieKey(rowChange2.getRowKey(), rowChange2.getPartitionPath()), rowChange2);
+ RawTripTestPayload rowChange3 = new RawTripTestPayload(recordStr3);
+ HoodieRecord record3 =
+ new HoodieRecord(new HoodieKey(rowChange3.getRowKey(), rowChange3.getPartitionPath()), rowChange3);
+ RawTripTestPayload rowChange4 = new RawTripTestPayload(recordStr4);
+ HoodieRecord record4 =
+ new HoodieRecord(new HoodieKey(rowChange4.getRowKey(), rowChange4.getPartitionPath()), rowChange4);
+
+ // We write record1, record2 to a parquet file, but the bloom filter contains (record1,
+ // record2, record3).
+ org.apache.hudi.common.bloom.BloomFilter filter = org.apache.hudi.common.bloom.BloomFilterFactory.createBloomFilter(10000, 0.0000001, -1, org.apache.hudi.common.bloom.BloomFilterTypeCode.SIMPLE.name());
+ filter.add(record3.getRecordKey());
+ org.apache.hudi.testutils.HoodieFlinkWriteableTestTable testTable = org.apache.hudi.testutils.HoodieFlinkWriteableTestTable.of(metaClient, SCHEMA, filter);
+ String fileId = testTable.addCommit("000").getFileIdWithInserts(partition, record1, record2);
+ String filename = testTable.getBaseFileNameById(fileId);
+
+ // The bloom filter contains 3 records
+ assertTrue(filter.mightContain(record1.getRecordKey()));
+ assertTrue(filter.mightContain(record2.getRecordKey()));
+ assertTrue(filter.mightContain(record3.getRecordKey()));
+ assertFalse(filter.mightContain(record4.getRecordKey()));
+
+ // Compare with file
+ java.util.List<String> uuids =
+ asList(record1.getRecordKey(), record2.getRecordKey(), record3.getRecordKey(), record4.getRecordKey());
+
+ org.apache.hudi.config.HoodieWriteConfig config = org.apache.hudi.config.HoodieWriteConfig.newBuilder().withPath(basePath).build();
+ HoodieFlinkTable table = HoodieFlinkTable.create(config, context, metaClient);
+ org.apache.hudi.io.HoodieKeyLookupHandle keyHandle = new org.apache.hudi.io.HoodieKeyLookupHandle<>(config, table, Pair.of(partition, fileId));
+ java.util.List<String> results = keyHandle.checkCandidatesAgainstFile(hadoopConf, uuids,
+ new org.apache.hadoop.fs.Path(java.nio.file.Paths.get(basePath, partition, filename).toString()));
+ assertEquals(results.size(), 2);
+ assertTrue(results.get(0).equals("1eb5b87a-1feh-4edd-87b4-6ec96dc405a0")
+ || results.get(1).equals("1eb5b87a-1feh-4edd-87b4-6ec96dc405a0"));
+ assertTrue(results.get(0).equals("2eb5b87b-1feu-4edd-87b4-6ec96dc405a0")
+ || results.get(1).equals("2eb5b87b-1feu-4edd-87b4-6ec96dc405a0"));
+ // TODO(vc): Need more coverage on actual filenames
+ // assertTrue(results.get(0)._2().equals(filename));
+ // assertTrue(results.get(1)._2().equals(filename));
+ }
+
+ @org.junit.jupiter.params.ParameterizedTest(name = TEST_NAME_WITH_PARAMS)
Review comment:
ditto
##########
File path: hudi-client/hudi-flink-client/src/test/java/org/apache/hudi/testutils/HoodieFlinkWriteableTestTable.java
##########
@@ -0,0 +1,143 @@
+/*
+ * 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.testutils;
+
+import static org.apache.hudi.common.testutils.FileCreateUtils.baseFileName;
+
+public class HoodieFlinkWriteableTestTable extends org.apache.hudi.common.testutils.HoodieTestTable {
+ private static final org.apache.log4j.Logger LOG = org.apache.log4j.LogManager.getLogger(org.apache.hudi.testutils.HoodieFlinkWriteableTestTable.class);
+
+ private final org.apache.avro.Schema schema;
+ private final org.apache.hudi.common.bloom.BloomFilter filter;
Review comment:
ditto, please move all the full package info to import
----------------------------------------------------------------
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.
For queries about this service, please contact Infrastructure at:
users@infra.apache.org