|
20 | 20 |
|
21 | 21 | import org.apache.hudi.avro.model.HoodieRollbackPartitionMetadata; |
22 | 22 | import org.apache.hudi.avro.model.HoodieRollbackPlan; |
| 23 | +import org.apache.hudi.avro.model.HoodieRollbackRequest; |
| 24 | +import org.apache.hudi.client.SparkRDDWriteClient; |
| 25 | +import org.apache.hudi.client.WriteStatus; |
| 26 | +import org.apache.hudi.client.common.HoodieSparkEngineContext; |
23 | 27 | import org.apache.hudi.common.HoodieRollbackStat; |
24 | 28 | import org.apache.hudi.common.model.FileSlice; |
25 | 29 | import org.apache.hudi.common.model.HoodieFileGroup; |
| 30 | +import org.apache.hudi.common.model.HoodieRecord; |
| 31 | +import org.apache.hudi.common.table.HoodieTableMetaClient; |
26 | 32 | import org.apache.hudi.common.table.timeline.HoodieInstant; |
27 | 33 | import org.apache.hudi.common.table.timeline.HoodieTimeline; |
| 34 | +import org.apache.hudi.common.testutils.HoodieTestDataGenerator; |
28 | 35 | import org.apache.hudi.common.testutils.HoodieTestTable; |
29 | 36 | import org.apache.hudi.config.HoodieWriteConfig; |
30 | 37 | import org.apache.hudi.table.HoodieTable; |
31 | 38 | import org.apache.hudi.table.marker.WriteMarkersFactory; |
| 39 | +import org.apache.hudi.testutils.Assertions; |
32 | 40 |
|
| 41 | +import org.apache.hadoop.fs.FileSystem; |
| 42 | +import org.apache.spark.api.java.JavaRDD; |
33 | 43 | import org.junit.jupiter.api.AfterEach; |
34 | 44 | import org.junit.jupiter.api.BeforeEach; |
35 | 45 | import org.junit.jupiter.api.Test; |
36 | 46 | import org.junit.jupiter.params.ParameterizedTest; |
37 | 47 | import org.junit.jupiter.params.provider.ValueSource; |
| 48 | +import org.mockito.Mockito; |
| 49 | +import org.mockito.MockitoAnnotations; |
38 | 50 |
|
39 | 51 | import java.io.IOException; |
40 | 52 | import java.util.ArrayList; |
| 53 | +import java.util.Arrays; |
41 | 54 | import java.util.Collections; |
42 | 55 | import java.util.List; |
43 | 56 | import java.util.Map; |
44 | 57 | import java.util.stream.Collectors; |
45 | 58 |
|
46 | 59 | import static org.apache.hudi.common.testutils.HoodieTestDataGenerator.DEFAULT_FIRST_PARTITION_PATH; |
47 | 60 | import static org.apache.hudi.common.testutils.HoodieTestDataGenerator.DEFAULT_SECOND_PARTITION_PATH; |
| 61 | +import static org.apache.hudi.common.testutils.HoodieTestDataGenerator.DEFAULT_THIRD_PARTITION_PATH; |
48 | 62 | import static org.junit.jupiter.api.Assertions.assertEquals; |
49 | 63 | import static org.junit.jupiter.api.Assertions.assertFalse; |
50 | 64 | import static org.junit.jupiter.api.Assertions.assertTrue; |
51 | 65 | import static org.junit.jupiter.api.Assertions.fail; |
| 66 | +import static org.mockito.ArgumentMatchers.any; |
52 | 67 |
|
53 | 68 | public class TestCopyOnWriteRollbackActionExecutor extends HoodieClientRollbackTestBase { |
54 | 69 | @BeforeEach |
@@ -133,6 +148,58 @@ public void testCopyOnWriteRollbackActionExecutorForFileListingAsGenerateFile() |
133 | 148 | assertFalse(testTable.baseFileExists(p2, "002", "id22")); |
134 | 149 | } |
135 | 150 |
|
| 151 | + @Test |
| 152 | + public void testListBasedRollbackStrategy() throws Exception { |
| 153 | + //just generate two partitions |
| 154 | + dataGen = new HoodieTestDataGenerator(new String[] {DEFAULT_FIRST_PARTITION_PATH, DEFAULT_SECOND_PARTITION_PATH, DEFAULT_THIRD_PARTITION_PATH}); |
| 155 | + HoodieWriteConfig cfg = getConfigBuilder().withRollbackUsingMarkers(false).build(); |
| 156 | + // 1. prepare data |
| 157 | + HoodieTestDataGenerator.writePartitionMetadataDeprecated(fs, new String[] {DEFAULT_FIRST_PARTITION_PATH, DEFAULT_SECOND_PARTITION_PATH}, basePath); |
| 158 | + SparkRDDWriteClient client = getHoodieWriteClient(cfg); |
| 159 | + |
| 160 | + String newCommitTime = "001"; |
| 161 | + client.startCommitWithTime(newCommitTime); |
| 162 | + List<HoodieRecord> records = dataGen.generateInsertsContainsAllPartitions(newCommitTime, 3); |
| 163 | + JavaRDD<HoodieRecord> writeRecords = jsc.parallelize(records, 1); |
| 164 | + JavaRDD<WriteStatus> statuses = client.upsert(writeRecords, newCommitTime); |
| 165 | + Assertions.assertNoWriteErrors(statuses.collect()); |
| 166 | + |
| 167 | + newCommitTime = "002"; |
| 168 | + client.startCommitWithTime(newCommitTime); |
| 169 | + records = dataGen.generateUpdates(newCommitTime, records); |
| 170 | + statuses = client.upsert(jsc.parallelize(records, 1), newCommitTime); |
| 171 | + Assertions.assertNoWriteErrors(statuses.collect()); |
| 172 | + |
| 173 | + context = new HoodieSparkEngineContext(jsc); |
| 174 | + metaClient = HoodieTableMetaClient.reload(metaClient); |
| 175 | + HoodieTable table = this.getHoodieTable(metaClient, cfg); |
| 176 | + HoodieInstant needRollBackInstant = new HoodieInstant(false, HoodieTimeline.COMMIT_ACTION, "002"); |
| 177 | + String rollbackInstant = "003"; |
| 178 | + |
| 179 | + ListingBasedRollbackStrategy rollbackStrategy = new ListingBasedRollbackStrategy(table, context, table.getConfig(), rollbackInstant); |
| 180 | + List<HoodieRollbackRequest> rollBackRequests = rollbackStrategy.getRollbackRequests(needRollBackInstant); |
| 181 | + rollBackRequests.forEach(entry -> System.out.println(" " + entry.getPartitionPath() + ", " + entry.getFileId() + " " |
| 182 | + + Arrays.toString(entry.getFilesToBeDeleted().toArray()))); |
| 183 | + |
| 184 | + HoodieRollbackRequest rollbackRequest = rollBackRequests.stream().filter(entry -> entry.getPartitionPath().equals(DEFAULT_FIRST_PARTITION_PATH)).findFirst().get(); |
| 185 | + |
| 186 | + FileSystem fs = Mockito.mock(FileSystem.class); |
| 187 | + MockitoAnnotations.initMocks(this); |
| 188 | + |
| 189 | + // mock to throw exception when fs.exists() is invoked |
| 190 | + System.out.println("Fs.exists() call for " + rollbackRequest.getFilesToBeDeleted().get(0).toString()); |
| 191 | + Mockito.when(fs.exists(any())) |
| 192 | + .thenThrow(new IOException("Failing exists call for " + rollbackRequest.getFilesToBeDeleted().get(0))); |
| 193 | + |
| 194 | + rollbackStrategy = new ListingBasedRollbackStrategy(table, context, cfg, rollbackInstant); |
| 195 | + List<HoodieRollbackRequest> rollBackRequestsUpdated = rollbackStrategy.getRollbackRequests(needRollBackInstant); |
| 196 | + rollBackRequestsUpdated.forEach(entry -> System.out.println(" " + entry.getPartitionPath() + ", " + entry.getFileId() + " " |
| 197 | + + Arrays.toString(entry.getFilesToBeDeleted().toArray()))); |
| 198 | + |
| 199 | + assertEquals(rollBackRequests, rollBackRequestsUpdated); |
| 200 | + } |
| 201 | + |
| 202 | + |
136 | 203 | // Verify that rollback works with replacecommit |
137 | 204 | @ParameterizedTest |
138 | 205 | @ValueSource(booleans = {true, false}) |
|
0 commit comments