1
0

[HUDI-2472] Fix few Cleaner tests with metadata table enabled (#3825)

This commit is contained in:
Manoj Govindassamy
2021-10-20 15:57:00 -07:00
committed by GitHub
parent e355ab52db
commit b68c5a68f9
6 changed files with 195 additions and 62 deletions

View File

@@ -63,6 +63,7 @@ import org.apache.hudi.common.table.timeline.versioning.clean.CleanMetadataMigra
import org.apache.hudi.common.table.timeline.versioning.clean.CleanPlanMigrator;
import org.apache.hudi.common.table.timeline.versioning.clean.CleanPlanV1MigrationHandler;
import org.apache.hudi.common.table.view.TableFileSystemView;
import org.apache.hudi.common.testutils.HoodieMetadataTestTable;
import org.apache.hudi.common.testutils.HoodieTestTable;
import org.apache.hudi.common.testutils.HoodieTestUtils;
import org.apache.hudi.common.util.CleanerUtils;
@@ -76,6 +77,8 @@ import org.apache.hudi.config.HoodieWriteConfig;
import org.apache.hudi.exception.HoodieIOException;
import org.apache.hudi.index.HoodieIndex;
import org.apache.hudi.index.SparkHoodieIndex;
import org.apache.hudi.metadata.HoodieTableMetadataWriter;
import org.apache.hudi.metadata.SparkHoodieBackedTableMetadataWriter;
import org.apache.hudi.table.action.clean.CleanPlanner;
import org.apache.hudi.testutils.HoodieClientTestBase;
import org.apache.log4j.LogManager;
@@ -635,29 +638,38 @@ public class TestCleaner extends HoodieClientTestBase {
}
/**
* Test HoodieTable.clean() Cleaning by versions logic.
* Test Hudi COW Table Cleaner - Keep the latest file versions policy.
*/
@ParameterizedTest
@ValueSource(booleans = {false, true})
public void testKeepLatestFileVersions(Boolean enableBootstrapSourceClean) throws Exception {
HoodieWriteConfig config =
HoodieWriteConfig.newBuilder().withPath(basePath)
.withMetadataConfig(HoodieMetadataConfig.newBuilder().withAssumeDatePartitioning(true).enable(false).build())
.withMetadataConfig(HoodieMetadataConfig.newBuilder().withAssumeDatePartitioning(true).enable(true).build())
.withCompactionConfig(HoodieCompactionConfig.newBuilder()
.withCleanBootstrapBaseFileEnabled(enableBootstrapSourceClean)
.withCleanerPolicy(HoodieCleaningPolicy.KEEP_LATEST_FILE_VERSIONS).retainFileVersions(1).build())
.build();
HoodieTestTable testTable = HoodieTestTable.of(metaClient);
String p0 = "2020/01/01";
String p1 = "2020/01/02";
Map<String, List<BootstrapFileMapping>> bootstrapMapping = enableBootstrapSourceClean ? generateBootstrapIndexAndSourceData(p0, p1) : null;
HoodieTableMetadataWriter metadataWriter = SparkHoodieBackedTableMetadataWriter.create(hadoopConf, config, context);
HoodieTestTable testTable = HoodieMetadataTestTable.of(metaClient, metadataWriter);
final String p0 = "2020/01/01";
final String p1 = "2020/01/02";
final Map<String, List<BootstrapFileMapping>> bootstrapMapping = enableBootstrapSourceClean
? generateBootstrapIndexAndSourceData(p0, p1) : null;
// make 1 commit, with 1 file per partition
String file1P0C0 = enableBootstrapSourceClean ? bootstrapMapping.get(p0).get(0).getFileId()
final String file1P0C0 = enableBootstrapSourceClean ? bootstrapMapping.get(p0).get(0).getFileId()
: UUID.randomUUID().toString();
String file1P1C0 = enableBootstrapSourceClean ? bootstrapMapping.get(p1).get(0).getFileId()
final String file1P1C0 = enableBootstrapSourceClean ? bootstrapMapping.get(p1).get(0).getFileId()
: UUID.randomUUID().toString();
testTable.addCommit("00000000000001").withBaseFilesInPartition(p0, file1P0C0).withBaseFilesInPartition(p1, file1P1C0);
Map<String, List<Pair<String, Integer>>> c1PartitionToFilesNameLengthMap = new HashMap<>();
c1PartitionToFilesNameLengthMap.put(p0, Collections.singletonList(Pair.of(file1P0C0, 100)));
c1PartitionToFilesNameLengthMap.put(p1, Collections.singletonList(Pair.of(file1P1C0, 200)));
testTable.doWriteOperation("00000000000001", WriteOperationType.INSERT, Arrays.asList(p0, p1),
c1PartitionToFilesNameLengthMap, false, false);
List<HoodieCleanStat> hoodieCleanStatsOne = runCleaner(config);
assertEquals(0, hoodieCleanStatsOne.size(), "Must not clean any files");
@@ -665,17 +677,21 @@ public class TestCleaner extends HoodieClientTestBase {
assertTrue(testTable.baseFileExists(p1, "00000000000001", file1P1C0));
// make next commit, with 1 insert & 1 update per partition
Map<String, String> partitionAndFileId002 = testTable.addCommit("00000000000002")
.withBaseFilesInPartition(p0, file1P0C0)
.withBaseFilesInPartition(p1, file1P1C0)
.getFileIdsWithBaseFilesInPartitions(p0, p1);
final String file2P0C1 = UUID.randomUUID().toString();
final String file2P1C1 = UUID.randomUUID().toString();
Map<String, List<Pair<String, Integer>>> c2PartitionToFilesNameLengthMap = new HashMap<>();
c2PartitionToFilesNameLengthMap.put(p0, Arrays.asList(Pair.of(file1P0C0, 101), Pair.of(file2P0C1, 100)));
c2PartitionToFilesNameLengthMap.put(p1, Arrays.asList(Pair.of(file1P1C0, 201), Pair.of(file2P1C1, 200)));
testTable.doWriteOperation("00000000000002", WriteOperationType.UPSERT, Collections.emptyList(),
c2PartitionToFilesNameLengthMap, false, false);
// enableBootstrapSourceClean would delete the bootstrap base file at the same time
List<HoodieCleanStat> hoodieCleanStatsTwo = runCleaner(config, 1);
// enableBootstrapSourceClean would delete the bootstrap base file as the same time
HoodieCleanStat cleanStat = getCleanStat(hoodieCleanStatsTwo, p0);
assertEquals(enableBootstrapSourceClean ? 2 : 1, cleanStat.getSuccessDeleteFiles().size()
+ (cleanStat.getSuccessDeleteBootstrapBaseFiles() == null ? 0
: cleanStat.getSuccessDeleteBootstrapBaseFiles().size()), "Must clean at least 1 file");
if (enableBootstrapSourceClean) {
HoodieFileStatus fstatus =
bootstrapMapping.get(p0).get(0).getBootstrapFileStatus();
@@ -686,9 +702,8 @@ public class TestCleaner extends HoodieClientTestBase {
assertFalse(Files.exists(Paths.get(bootstrapMapping.get(
p0).get(0).getBootstrapFileStatus().getPath().getUri())));
}
cleanStat = getCleanStat(hoodieCleanStatsTwo, p1);
String file2P0C1 = partitionAndFileId002.get(p0);
String file2P1C1 = partitionAndFileId002.get(p1);
assertTrue(testTable.baseFileExists(p0, "00000000000002", file2P0C1));
assertTrue(testTable.baseFileExists(p1, "00000000000002", file2P1C1));
assertFalse(testTable.baseFileExists(p0, "00000000000001", file1P0C0));
@@ -696,6 +711,7 @@ public class TestCleaner extends HoodieClientTestBase {
assertEquals(enableBootstrapSourceClean ? 2 : 1, cleanStat.getSuccessDeleteFiles().size()
+ (cleanStat.getSuccessDeleteBootstrapBaseFiles() == null ? 0
: cleanStat.getSuccessDeleteBootstrapBaseFiles().size()), "Must clean at least 1 file");
if (enableBootstrapSourceClean) {
HoodieFileStatus fstatus =
bootstrapMapping.get(p1).get(0).getBootstrapFileStatus();
@@ -708,9 +724,13 @@ public class TestCleaner extends HoodieClientTestBase {
}
// make next commit, with 2 updates to existing files, and 1 insert
String file3P0C2 = testTable.addCommit("00000000000003")
.withBaseFilesInPartition(p0, file1P0C0, file2P0C1)
.getFileIdsWithBaseFilesInPartitions(p0).get(p0);
final String file3P0C2 = UUID.randomUUID().toString();
Map<String, List<Pair<String, Integer>>> c3PartitionToFilesNameLengthMap = new HashMap<>();
c3PartitionToFilesNameLengthMap.put(p0, Arrays.asList(Pair.of(file1P0C0, 102), Pair.of(file2P0C1, 101),
Pair.of(file3P0C2, 100)));
testTable.doWriteOperation("00000000000003", WriteOperationType.UPSERT, Collections.emptyList(),
c3PartitionToFilesNameLengthMap, false, false);
List<HoodieCleanStat> hoodieCleanStatsThree = runCleaner(config, 3);
assertEquals(2,
getCleanStat(hoodieCleanStatsThree, p0)
@@ -721,6 +741,7 @@ public class TestCleaner extends HoodieClientTestBase {
// No cleaning on partially written file, with no commit.
testTable.forCommit("00000000000004").withBaseFilesInPartition(p0, file3P0C2);
List<HoodieCleanStat> hoodieCleanStatsFour = runCleaner(config);
assertEquals(0, hoodieCleanStatsFour.size(), "Must not clean any files");
assertTrue(testTable.baseFileExists(p0, "00000000000003", file3P0C2));