/** * Legacy indices without lucene CRC32 did never write or calculate checksums for segments_N files * but for other files */ @Test public void testRecoveryDiffWithLegacyCommit() { Map<String, StoreFileMetaData> metaDataMap = new HashMap<>(); metaDataMap.put( "segments_1", new StoreFileMetaData("segments_1", 50, null, null, new BytesRef(new byte[] {1}))); metaDataMap.put( "_0_1.del", new StoreFileMetaData("_0_1.del", 42, "foobarbaz", null, new BytesRef())); Store.MetadataSnapshot first = new Store.MetadataSnapshot(metaDataMap); Store.MetadataSnapshot second = new Store.MetadataSnapshot(metaDataMap); Store.RecoveryDiff recoveryDiff = first.recoveryDiff(second); assertEquals(recoveryDiff.toString(), recoveryDiff.different.size(), 2); }
public static void assertConsistent(Store store, Store.MetadataSnapshot metadata) throws IOException { for (String file : store.directory().listAll()) { if (!IndexWriter.WRITE_LOCK_NAME.equals(file) && !IndexFileNames.OLD_SEGMENTS_GEN.equals(file) && !Store.isChecksum(file)) { assertTrue( file + " is not in the map: " + metadata.asMap().size() + " vs. " + store.directory().listAll().length, metadata.asMap().containsKey(file)); } else { assertFalse( file + " is not in the map: " + metadata.asMap().size() + " vs. " + store.directory().listAll().length, metadata.asMap().containsKey(file)); } } }
@Test public void testRecoveryDiff() throws IOException, InterruptedException { int numDocs = 2 + random().nextInt(100); List<Document> docs = new ArrayList<>(); for (int i = 0; i < numDocs; i++) { Document doc = new Document(); doc.add( new StringField("id", "" + i, random().nextBoolean() ? Field.Store.YES : Field.Store.NO)); doc.add( new TextField( "body", TestUtil.randomRealisticUnicodeString(random()), random().nextBoolean() ? Field.Store.YES : Field.Store.NO)); doc.add( new SortedDocValuesField( "dv", new BytesRef(TestUtil.randomRealisticUnicodeString(random())))); docs.add(doc); } long seed = random().nextLong(); Store.MetadataSnapshot first; { Random random = new Random(seed); IndexWriterConfig iwc = new IndexWriterConfig(new MockAnalyzer(random)).setCodec(actualDefaultCodec()); iwc.setMergePolicy(NoMergePolicy.INSTANCE); iwc.setUseCompoundFile(random.nextBoolean()); iwc.setMaxThreadStates(1); final ShardId shardId = new ShardId(new Index("index"), 1); DirectoryService directoryService = new LuceneManagedDirectoryService(random); Store store = new Store( shardId, ImmutableSettings.EMPTY, directoryService, randomDistributor(random, directoryService), new DummyShardLock(shardId)); IndexWriter writer = new IndexWriter(store.directory(), iwc); final boolean lotsOfSegments = rarely(random); for (Document d : docs) { writer.addDocument(d); if (lotsOfSegments && random.nextBoolean()) { writer.commit(); } else if (rarely(random)) { writer.commit(); } } writer.commit(); writer.close(); first = store.getMetadata(); assertDeleteContent(store, directoryService); store.close(); } long time = new Date().getTime(); while (time == new Date().getTime()) { Thread.sleep(10); // bump the time } Store.MetadataSnapshot second; Store store; { Random random = new Random(seed); IndexWriterConfig iwc = new IndexWriterConfig(new MockAnalyzer(random)).setCodec(actualDefaultCodec()); iwc.setMergePolicy(NoMergePolicy.INSTANCE); iwc.setUseCompoundFile(random.nextBoolean()); iwc.setMaxThreadStates(1); final ShardId shardId = new ShardId(new Index("index"), 1); DirectoryService directoryService = new LuceneManagedDirectoryService(random); store = new Store( shardId, ImmutableSettings.EMPTY, directoryService, randomDistributor(random, directoryService), new DummyShardLock(shardId)); IndexWriter writer = new IndexWriter(store.directory(), iwc); final boolean lotsOfSegments = rarely(random); for (Document d : docs) { writer.addDocument(d); if (lotsOfSegments && random.nextBoolean()) { writer.commit(); } else if (rarely(random)) { writer.commit(); } } writer.commit(); writer.close(); second = store.getMetadata(); } Store.RecoveryDiff diff = first.recoveryDiff(second); assertThat(first.size(), equalTo(second.size())); for (StoreFileMetaData md : first) { assertThat(second.get(md.name()), notNullValue()); // si files are different - containing timestamps etc assertThat(second.get(md.name()).isSame(md), equalTo(false)); } assertThat(diff.different.size(), equalTo(first.size())); assertThat( diff.identical.size(), equalTo(0)); // in lucene 5 nothing is identical - we use random ids in file headers assertThat(diff.missing, empty()); // check the self diff Store.RecoveryDiff selfDiff = first.recoveryDiff(first); assertThat(selfDiff.identical.size(), equalTo(first.size())); assertThat(selfDiff.different, empty()); assertThat(selfDiff.missing, empty()); // lets add some deletes Random random = new Random(seed); IndexWriterConfig iwc = new IndexWriterConfig(new MockAnalyzer(random)).setCodec(actualDefaultCodec()); iwc.setMergePolicy(NoMergePolicy.INSTANCE); iwc.setUseCompoundFile(random.nextBoolean()); iwc.setMaxThreadStates(1); iwc.setOpenMode(IndexWriterConfig.OpenMode.APPEND); IndexWriter writer = new IndexWriter(store.directory(), iwc); writer.deleteDocuments(new Term("id", Integer.toString(random().nextInt(numDocs)))); writer.commit(); writer.close(); Store.MetadataSnapshot metadata = store.getMetadata(); StoreFileMetaData delFile = null; for (StoreFileMetaData md : metadata) { if (md.name().endsWith(".liv")) { delFile = md; break; } } Store.RecoveryDiff afterDeleteDiff = metadata.recoveryDiff(second); if (delFile != null) { assertThat( afterDeleteDiff.identical.size(), equalTo(metadata.size() - 2)); // segments_N + del file assertThat(afterDeleteDiff.different.size(), equalTo(0)); assertThat(afterDeleteDiff.missing.size(), equalTo(2)); } else { // an entire segment must be missing (single doc segment got dropped) assertThat(afterDeleteDiff.identical.size(), greaterThan(0)); assertThat(afterDeleteDiff.different.size(), equalTo(0)); assertThat(afterDeleteDiff.missing.size(), equalTo(1)); // the commit file is different } // check the self diff selfDiff = metadata.recoveryDiff(metadata); assertThat(selfDiff.identical.size(), equalTo(metadata.size())); assertThat(selfDiff.different, empty()); assertThat(selfDiff.missing, empty()); // add a new commit iwc = new IndexWriterConfig(new MockAnalyzer(random)).setCodec(actualDefaultCodec()); iwc.setMergePolicy(NoMergePolicy.INSTANCE); iwc.setUseCompoundFile( true); // force CFS - easier to test here since we know it will add 3 files iwc.setMaxThreadStates(1); iwc.setOpenMode(IndexWriterConfig.OpenMode.APPEND); writer = new IndexWriter(store.directory(), iwc); writer.addDocument(docs.get(0)); writer.close(); Store.MetadataSnapshot newCommitMetaData = store.getMetadata(); Store.RecoveryDiff newCommitDiff = newCommitMetaData.recoveryDiff(metadata); if (delFile != null) { assertThat( newCommitDiff.identical.size(), equalTo( newCommitMetaData.size() - 5)); // segments_N, del file, cfs, cfe, si for the new segment assertThat(newCommitDiff.different.size(), equalTo(1)); // the del file must be different assertThat(newCommitDiff.different.get(0).name(), endsWith(".liv")); assertThat( newCommitDiff.missing.size(), equalTo(4)); // segments_N,cfs, cfe, si for the new segment } else { assertThat( newCommitDiff.identical.size(), equalTo(newCommitMetaData.size() - 4)); // segments_N, cfs, cfe, si for the new segment assertThat(newCommitDiff.different.size(), equalTo(0)); assertThat( newCommitDiff.missing.size(), equalTo( 4)); // an entire segment must be missing (single doc segment got dropped) plus the // commit is different } store.deleteContent(); IOUtils.close(store); }
@Test public void testMixedChecksums() throws IOException { final ShardId shardId = new ShardId(new Index("index"), 1); DirectoryService directoryService = new LuceneManagedDirectoryService(random()); Store store = new Store( shardId, ImmutableSettings.EMPTY, directoryService, randomDistributor(directoryService), new DummyShardLock(shardId)); // this time random codec.... IndexWriter writer = new IndexWriter( store.directory(), newIndexWriterConfig(random(), new MockAnalyzer(random())) .setCodec(actualDefaultCodec())); int docs = 1 + random().nextInt(100); for (int i = 0; i < docs; i++) { Document doc = new Document(); doc.add( new TextField("id", "" + i, random().nextBoolean() ? Field.Store.YES : Field.Store.NO)); doc.add( new TextField( "body", TestUtil.randomRealisticUnicodeString(random()), random().nextBoolean() ? Field.Store.YES : Field.Store.NO)); doc.add( new SortedDocValuesField( "dv", new BytesRef(TestUtil.randomRealisticUnicodeString(random())))); writer.addDocument(doc); } if (random().nextBoolean()) { for (int i = 0; i < docs; i++) { if (random().nextBoolean()) { Document doc = new Document(); doc.add( new TextField( "id", "" + i, random().nextBoolean() ? Field.Store.YES : Field.Store.NO)); doc.add( new TextField( "body", TestUtil.randomRealisticUnicodeString(random()), random().nextBoolean() ? Field.Store.YES : Field.Store.NO)); writer.updateDocument(new Term("id", "" + i), doc); } } } if (random().nextBoolean()) { DirectoryReader.open(writer, random().nextBoolean()).close(); // flush } Store.MetadataSnapshot metadata; // check before we committed try { store.getMetadata(); fail("no index present - expected exception"); } catch (IndexNotFoundException ex) { // expected } assertThat(store.getMetadataOrEmpty(), is(Store.MetadataSnapshot.EMPTY)); // nothing committed writer.commit(); writer.close(); Store.LegacyChecksums checksums = new Store.LegacyChecksums(); metadata = store.getMetadata(); assertThat(metadata.asMap().isEmpty(), is(false)); for (StoreFileMetaData meta : metadata) { try (IndexInput input = store.directory().openInput(meta.name(), IOContext.DEFAULT)) { if (meta.checksum() == null) { String checksum = null; try { CodecUtil.retrieveChecksum(input); fail("expected a corrupt index - posting format has not checksums"); } catch (CorruptIndexException | IndexFormatTooOldException | IndexFormatTooNewException ex) { try (ChecksumIndexInput checksumIndexInput = store.directory().openChecksumInput(meta.name(), IOContext.DEFAULT)) { checksumIndexInput.seek(meta.length()); checksum = Store.digestToString(checksumIndexInput.getChecksum()); } // fine - it's a postings format without checksums checksums.add(new StoreFileMetaData(meta.name(), meta.length(), checksum, null)); } } else { String checksum = Store.digestToString(CodecUtil.retrieveChecksum(input)); assertThat( "File: " + meta.name() + " has a different checksum", meta.checksum(), equalTo(checksum)); assertThat(meta.hasLegacyChecksum(), equalTo(false)); assertThat(meta.writtenBy(), equalTo(Version.LATEST)); } } } assertConsistent(store, metadata); checksums.write(store); metadata = store.getMetadata(); assertThat(metadata.asMap().isEmpty(), is(false)); for (StoreFileMetaData meta : metadata) { assertThat( "file: " + meta.name() + " has a null checksum", meta.checksum(), not(nullValue())); if (meta.hasLegacyChecksum()) { try (ChecksumIndexInput checksumIndexInput = store.directory().openChecksumInput(meta.name(), IOContext.DEFAULT)) { checksumIndexInput.seek(meta.length()); assertThat( meta.checksum(), equalTo(Store.digestToString(checksumIndexInput.getChecksum()))); } } else { try (IndexInput input = store.directory().openInput(meta.name(), IOContext.DEFAULT)) { String checksum = Store.digestToString(CodecUtil.retrieveChecksum(input)); assertThat( "File: " + meta.name() + " has a different checksum", meta.checksum(), equalTo(checksum)); assertThat(meta.hasLegacyChecksum(), equalTo(false)); assertThat(meta.writtenBy(), equalTo(Version.LATEST)); } } } assertConsistent(store, metadata); TestUtil.checkIndex(store.directory()); assertDeleteContent(store, directoryService); IOUtils.close(store); }
@Test public void testNewChecksums() throws IOException { final ShardId shardId = new ShardId(new Index("index"), 1); DirectoryService directoryService = new LuceneManagedDirectoryService(random()); Store store = new Store( shardId, ImmutableSettings.EMPTY, directoryService, randomDistributor(directoryService), new DummyShardLock(shardId)); // set default codec - all segments need checksums IndexWriter writer = new IndexWriter( store.directory(), newIndexWriterConfig(random(), new MockAnalyzer(random())) .setCodec(actualDefaultCodec())); int docs = 1 + random().nextInt(100); for (int i = 0; i < docs; i++) { Document doc = new Document(); doc.add( new TextField("id", "" + i, random().nextBoolean() ? Field.Store.YES : Field.Store.NO)); doc.add( new TextField( "body", TestUtil.randomRealisticUnicodeString(random()), random().nextBoolean() ? Field.Store.YES : Field.Store.NO)); doc.add( new SortedDocValuesField( "dv", new BytesRef(TestUtil.randomRealisticUnicodeString(random())))); writer.addDocument(doc); } if (random().nextBoolean()) { for (int i = 0; i < docs; i++) { if (random().nextBoolean()) { Document doc = new Document(); doc.add( new TextField( "id", "" + i, random().nextBoolean() ? Field.Store.YES : Field.Store.NO)); doc.add( new TextField( "body", TestUtil.randomRealisticUnicodeString(random()), random().nextBoolean() ? Field.Store.YES : Field.Store.NO)); writer.updateDocument(new Term("id", "" + i), doc); } } } if (random().nextBoolean()) { DirectoryReader.open(writer, random().nextBoolean()).close(); // flush } Store.MetadataSnapshot metadata; // check before we committed try { store.getMetadata(); fail("no index present - expected exception"); } catch (IndexNotFoundException ex) { // expected } assertThat(store.getMetadataOrEmpty(), is(Store.MetadataSnapshot.EMPTY)); // nothing committed writer.commit(); writer.close(); metadata = store.getMetadata(); assertThat(metadata.asMap().isEmpty(), is(false)); for (StoreFileMetaData meta : metadata) { try (IndexInput input = store.directory().openInput(meta.name(), IOContext.DEFAULT)) { String checksum = Store.digestToString(CodecUtil.retrieveChecksum(input)); assertThat( "File: " + meta.name() + " has a different checksum", meta.checksum(), equalTo(checksum)); assertThat(meta.hasLegacyChecksum(), equalTo(false)); assertThat(meta.writtenBy(), equalTo(Version.LATEST)); if (meta.name().endsWith(".si") || meta.name().startsWith("segments_")) { assertThat(meta.hash().length, greaterThan(0)); } } } assertConsistent(store, metadata); TestUtil.checkIndex(store.directory()); assertDeleteContent(store, directoryService); IOUtils.close(store); }
// IF THIS TEST FAILS ON UPGRADE GO LOOK AT THE // OldSIMockingCodec!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! @Test public void testWriteLegacyChecksums() throws IOException { final ShardId shardId = new ShardId(new Index("index"), 1); DirectoryService directoryService = new LuceneManagedDirectoryService(random()); Store store = new Store( shardId, ImmutableSettings.EMPTY, directoryService, randomDistributor(directoryService), new DummyShardLock(shardId)); // set default codec - all segments need checksums final boolean usesOldCodec = randomBoolean(); IndexWriter writer = new IndexWriter( store.directory(), newIndexWriterConfig(random(), new MockAnalyzer(random())) .setCodec(usesOldCodec ? new OldSIMockingCodec() : actualDefaultCodec())); int docs = 1 + random().nextInt(100); for (int i = 0; i < docs; i++) { Document doc = new Document(); doc.add( new TextField("id", "" + i, random().nextBoolean() ? Field.Store.YES : Field.Store.NO)); doc.add( new TextField( "body", TestUtil.randomRealisticUnicodeString(random()), random().nextBoolean() ? Field.Store.YES : Field.Store.NO)); doc.add( new SortedDocValuesField( "dv", new BytesRef(TestUtil.randomRealisticUnicodeString(random())))); writer.addDocument(doc); } if (random().nextBoolean()) { for (int i = 0; i < docs; i++) { if (random().nextBoolean()) { Document doc = new Document(); doc.add( new TextField( "id", "" + i, random().nextBoolean() ? Field.Store.YES : Field.Store.NO)); doc.add( new TextField( "body", TestUtil.randomRealisticUnicodeString(random()), random().nextBoolean() ? Field.Store.YES : Field.Store.NO)); writer.updateDocument(new Term("id", "" + i), doc); } } } if (random().nextBoolean()) { DirectoryReader.open(writer, random().nextBoolean()).close(); // flush } Store.MetadataSnapshot metadata; // check before we committed try { store.getMetadata(); fail("no index present - expected exception"); } catch (IndexNotFoundException ex) { // expected } assertThat(store.getMetadataOrEmpty(), is(Store.MetadataSnapshot.EMPTY)); // nothing committed writer.close(); Store.LegacyChecksums checksums = new Store.LegacyChecksums(); Map<String, StoreFileMetaData> legacyMeta = new HashMap<>(); for (String file : store.directory().listAll()) { if (file.equals("write.lock") || file.equals(IndexFileNames.OLD_SEGMENTS_GEN)) { continue; } BytesRef hash = new BytesRef(); if (file.startsWith("segments")) { hash = Store.MetadataSnapshot.hashFile(store.directory(), file); } StoreFileMetaData storeFileMetaData = new StoreFileMetaData( file, store.directory().fileLength(file), file + "checksum", null, hash); legacyMeta.put(file, storeFileMetaData); checksums.add(storeFileMetaData); } checksums.write(store); metadata = store.getMetadata(); Map<String, StoreFileMetaData> stringStoreFileMetaDataMap = metadata.asMap(); assertThat(legacyMeta.size(), equalTo(stringStoreFileMetaDataMap.size())); if (usesOldCodec) { for (StoreFileMetaData meta : legacyMeta.values()) { assertTrue(meta.toString(), stringStoreFileMetaDataMap.containsKey(meta.name())); assertEquals(meta.name() + "checksum", meta.checksum()); assertTrue( meta + " vs. " + stringStoreFileMetaDataMap.get(meta.name()), stringStoreFileMetaDataMap.get(meta.name()).isSame(meta)); } } else { // even if we have a legacy checksum - if we use a new codec we should reuse for (StoreFileMetaData meta : legacyMeta.values()) { assertTrue(meta.toString(), stringStoreFileMetaDataMap.containsKey(meta.name())); assertFalse( meta + " vs. " + stringStoreFileMetaDataMap.get(meta.name()), stringStoreFileMetaDataMap.get(meta.name()).isSame(meta)); StoreFileMetaData storeFileMetaData = metadata.get(meta.name()); try (IndexInput input = store.openVerifyingInput(meta.name(), IOContext.DEFAULT, storeFileMetaData)) { assertTrue(storeFileMetaData.toString(), input instanceof Store.VerifyingIndexInput); input.seek(meta.length()); Store.verify(input); } } } assertDeleteContent(store, directoryService); IOUtils.close(store); }
@Test public void testCleanupFromSnapshot() throws IOException { final ShardId shardId = new ShardId(new Index("index"), 1); DirectoryService directoryService = new LuceneManagedDirectoryService(random()); Store store = new Store( shardId, ImmutableSettings.EMPTY, directoryService, randomDistributor(directoryService), new DummyShardLock(shardId)); // this time random codec.... IndexWriterConfig indexWriterConfig = newIndexWriterConfig(random(), new MockAnalyzer(random())).setCodec(actualDefaultCodec()); // we keep all commits and that allows us clean based on multiple snapshots indexWriterConfig.setIndexDeletionPolicy(NoDeletionPolicy.INSTANCE); IndexWriter writer = new IndexWriter(store.directory(), indexWriterConfig); int docs = 1 + random().nextInt(100); int numCommits = 0; for (int i = 0; i < docs; i++) { if (i > 0 && randomIntBetween(0, 10) == 0) { writer.commit(); numCommits++; } Document doc = new Document(); doc.add( new TextField("id", "" + i, random().nextBoolean() ? Field.Store.YES : Field.Store.NO)); doc.add( new TextField( "body", TestUtil.randomRealisticUnicodeString(random()), random().nextBoolean() ? Field.Store.YES : Field.Store.NO)); doc.add( new SortedDocValuesField( "dv", new BytesRef(TestUtil.randomRealisticUnicodeString(random())))); writer.addDocument(doc); } if (numCommits < 1) { writer.commit(); Document doc = new Document(); doc.add( new TextField( "id", "" + docs++, random().nextBoolean() ? Field.Store.YES : Field.Store.NO)); doc.add( new TextField( "body", TestUtil.randomRealisticUnicodeString(random()), random().nextBoolean() ? Field.Store.YES : Field.Store.NO)); doc.add( new SortedDocValuesField( "dv", new BytesRef(TestUtil.randomRealisticUnicodeString(random())))); writer.addDocument(doc); } Store.MetadataSnapshot firstMeta = store.getMetadata(); if (random().nextBoolean()) { for (int i = 0; i < docs; i++) { if (random().nextBoolean()) { Document doc = new Document(); doc.add( new TextField( "id", "" + i, random().nextBoolean() ? Field.Store.YES : Field.Store.NO)); doc.add( new TextField( "body", TestUtil.randomRealisticUnicodeString(random()), random().nextBoolean() ? Field.Store.YES : Field.Store.NO)); writer.updateDocument(new Term("id", "" + i), doc); } } } writer.commit(); writer.close(); Store.MetadataSnapshot secondMeta = store.getMetadata(); Store.LegacyChecksums checksums = new Store.LegacyChecksums(); Map<String, StoreFileMetaData> legacyMeta = new HashMap<>(); for (String file : store.directory().listAll()) { if (file.equals("write.lock") || file.equals(IndexFileNames.OLD_SEGMENTS_GEN)) { continue; } BytesRef hash = new BytesRef(); if (file.startsWith("segments")) { hash = Store.MetadataSnapshot.hashFile(store.directory(), file); } StoreFileMetaData storeFileMetaData = new StoreFileMetaData( file, store.directory().fileLength(file), file + "checksum", null, hash); legacyMeta.put(file, storeFileMetaData); checksums.add(storeFileMetaData); } checksums.write( store); // write one checksum file here - we expect it to survive all the cleanups if (randomBoolean()) { store.cleanupAndVerify("test", firstMeta); String[] strings = store.directory().listAll(); int numChecksums = 0; int numNotFound = 0; for (String file : strings) { assertTrue(firstMeta.contains(file) || Store.isChecksum(file)); if (Store.isChecksum(file)) { numChecksums++; } else if (secondMeta.contains(file) == false) { numNotFound++; } } assertTrue( "at least one file must not be in here since we have two commits?", numNotFound > 0); assertEquals( "we wrote one checksum but it's gone now? - checksums are supposed to be kept", numChecksums, 1); } else { store.cleanupAndVerify("test", secondMeta); String[] strings = store.directory().listAll(); int numChecksums = 0; int numNotFound = 0; for (String file : strings) { assertTrue(secondMeta.contains(file) || Store.isChecksum(file)); if (Store.isChecksum(file)) { numChecksums++; } else if (firstMeta.contains(file) == false) { numNotFound++; } } assertTrue( "at least one file must not be in here since we have two commits?", numNotFound > 0); assertEquals( "we wrote one checksum but it's gone now? - checksums are supposed to be kept", numChecksums, 1); } store.deleteContent(); IOUtils.close(store); }