List of usage examples for org.apache.lucene.index IndexWriterConfig setCodec
public IndexWriterConfig setCodec(Codec codec)
From source file:perf.AutoPrefixPerf.java
License:Apache License
public static void main(String[] args) throws Exception { String numbersFile = args[0]; String queriesFile = args[1]; Path indexPath = Paths.get(args[2]); int precStep = Integer.parseInt(args[3]); boolean useNumericField = (precStep != 0); int maxTermsInPrefix; int minTermsInPrefix; if (useNumericField == false) { minTermsInPrefix = Integer.parseInt(args[4]); maxTermsInPrefix = Integer.parseInt(args[5]); } else {//from w ww . j a va2 s . c om minTermsInPrefix = 0; maxTermsInPrefix = 0; } BytesRefBuilder binaryToken = new BytesRefBuilder(); binaryToken.grow(8); binaryToken.setLength(8); Directory dir = FSDirectory.open(indexPath); if (Files.notExists(indexPath) == false) { IndexWriterConfig iwc = new IndexWriterConfig(new StandardAnalyzer()); iwc.setMaxBufferedDocs(30000); iwc.setRAMBufferSizeMB(-1); iwc.setMergePolicy(new LogDocMergePolicy()); final PostingsFormat pf; if (useNumericField) { // Disable auto-prefix when testing NumericField! if (minTermsInPrefix != 0) { throw new IllegalArgumentException("only precStep or minTermsInPrefix should be non-zero"); } pf = new Lucene50PostingsFormat(25, 48, 0, 0); } else { /* if (minTermsInPrefix == 0) { throw new IllegalArgumentException("one of precStep or minTermsInPrefix must be non-zero"); } */ pf = new Lucene50PostingsFormat(25, 48, minTermsInPrefix, maxTermsInPrefix); //pf = new Lucene50PostingsFormat(25, 48, minTermsInPrefix, Integer.MAX_VALUE); } iwc.setCodec(new Lucene53Codec() { @Override public PostingsFormat getPostingsFormatForField(String field) { return pf; } }); iwc.setInfoStream(new PrintStreamInfoStream(System.out)); iwc.setMergeScheduler(new SerialMergeScheduler()); //TieredMergePolicy tmp = (TieredMergePolicy) iwc.getMergePolicy(); //tmp.setFloorSegmentMB(.1); //ConcurrentMergeScheduler cms = (ConcurrentMergeScheduler) iwc.getMergeScheduler(); // More concurrency (for SSD) //cms.setMaxMergesAndThreads(5, 3); final IndexWriter w = new IndexWriter(dir, iwc); Document doc = new Document(); Field field; if (useNumericField) { FieldType longFieldType = new FieldType(LongField.TYPE_NOT_STORED); longFieldType.setNumericPrecisionStep(precStep); longFieldType.freeze(); field = new LongField("number", 0L, longFieldType); doc.add(field); } else { FieldType longFieldType = new FieldType(TextField.TYPE_NOT_STORED); longFieldType.setIndexOptions(IndexOptions.DOCS_ONLY); longFieldType.setOmitNorms(true); longFieldType.setIndexRanges(true); longFieldType.freeze(); field = new Field("number", new BinaryTokenStream(binaryToken.get()), longFieldType); doc.add(field); } long startMS = System.currentTimeMillis(); // 64K buffer: InputStream is = new FileInputStream(numbersFile); BufferedReader reader = new BufferedReader(new InputStreamReader(is, "UTF-8"), 1 << 16); int count = 0; while (true) { String line = reader.readLine(); if (line == null) { break; } Long v = Long.parseLong(line.trim()); if (useNumericField) { field.setLongValue(v); } else { //NumericUtils.longToPrefixCoded(v, 0, binaryToken); longToBytes(v, binaryToken); //if (bytesToLong(binaryToken.get()) != v) { // throw new RuntimeException("wrong long: v=" + v + " vs " + bytesToLong(binaryToken.get())); //} } w.addDocument(doc); count++; if (count % 200000 == 0) { long ms = System.currentTimeMillis(); System.out.println("Indexed " + count + ": " + ((ms - startMS) / 1000.0) + " sec"); } } reader.close(); System.out.println( "Final Indexed " + count + ": " + ((System.currentTimeMillis() - startMS) / 1000.0) + " sec"); // nocommit just to make debugging easier: //System.out.println("Optimize..."); //w.forceMerge(1); System.out.println("Close..."); w.close(); System.out.println("After close: " + ((System.currentTimeMillis() - startMS) / 1000.0) + " sec"); // Print CheckIndex: ByteArrayOutputStream bos = new ByteArrayOutputStream(1024); CheckIndex checker = new CheckIndex(dir); checker.setInfoStream(new PrintStream(bos, false, IOUtils.UTF_8), true); CheckIndex.Status status = checker.checkIndex(); System.out.println("Done CheckIndex:"); System.out.println(bos.toString(IOUtils.UTF_8)); if (status.clean == false) { throw new IllegalStateException("CheckIndex failed"); } SegmentInfos infos = new SegmentInfos(); infos.read(dir); long totBytes = 0; for (SegmentCommitInfo info : infos) { totBytes += info.sizeInBytes(); } System.out.println("\nTotal index size: " + totBytes + " bytes"); } else { System.out.println("Skip indexing: index already exists"); } List<Query> queries = new ArrayList<>(); InputStream is = new FileInputStream(queriesFile); BufferedReader reader = new BufferedReader(new InputStreamReader(is, "UTF-8"), 1 << 16); while (true) { String line = reader.readLine(); if (line == null) { break; } String[] numbers = line.trim().split(" "); if (numbers.length != 2) { throw new IllegalArgumentException("could not parse query line: " + line); } long minValue = Long.parseLong(numbers[0]); long maxValue = Long.parseLong(numbers[1]); if (useNumericField) { queries.add(NumericRangeQuery.newLongRange("number", precStep, minValue, maxValue, true, true)); } else { longToBytes(minValue, binaryToken); BytesRef minTerm = binaryToken.toBytesRef(); longToBytes(maxValue, binaryToken); BytesRef maxTerm = binaryToken.toBytesRef(); queries.add(new TermRangeQuery("number", minTerm, maxTerm, true, true)); } if (queries.size() == 200) { break; } } DirectoryReader r = DirectoryReader.open(dir); IndexSearcher s = new IndexSearcher(r); s.setQueryCache(null); // don't bench the cache printQueryTerms((MultiTermQuery) queries.get(0), s); long bestMS = Long.MAX_VALUE; for (int iter = 0; iter < 10; iter++) { long startMS = System.currentTimeMillis(); long totalHits = 0; long hash = 0; for (Query query : queries) { TopDocs hits = s.search(query, 10); totalHits += hits.totalHits; hash = hash * 31 + hits.totalHits; } long ms = System.currentTimeMillis() - startMS; System.out.println("iter " + iter + ": " + ms + " msec; totalHits=" + totalHits + " hash=" + hash); if (ms < bestMS) { System.out.println(" **"); bestMS = ms; } } /* long t0 = System.currentTimeMillis(); long bytesUsed = 0; for(int i=0;i<1000;i++) { for(AtomicReaderContext ctx : r.leaves()) { bytesUsed += ((SegmentReader) ctx.reader()).ramBytesUsed(); } } System.out.println((System.currentTimeMillis() - t0) + " msec for 1000 ramBytesUsed: " + (bytesUsed / 1000)); */ r.close(); dir.close(); }
From source file:perf.DiskUsage.java
License:Apache License
public static void main(String args[]) throws Exception { if (args.length != 1) { System.err.println(// w ww.ja v a 2s .c om "java [-Djava.io.tmpdir=/scratch] [-Dmode=BEST_COMPRESSION] -cp lucene-core.jar:./build DiskUsage <path to lucene index>"); System.exit(1); } IndexWriterConfig conf = new IndexWriterConfig(null); conf.setOpenMode(OpenMode.CREATE); // force codec to write per-field filenames. conf.setCodec(new Lucene60Codec(Mode.valueOf(System.getProperty("mode", "BEST_SPEED"))) { @Override public PostingsFormat getPostingsFormatForField(String field) { return new Lucene50PostingsFormat(); } @Override public DocValuesFormat getDocValuesFormatForField(String field) { return new Lucene54DocValuesFormat(); } }); Path tmp = Files.createTempDirectory(null); System.err.println("analyzing... (using " + tmp + " for temporary storage)"); try (Directory dir = FSDirectory.open(Paths.get(args[0])); DirectoryReader reader = DirectoryReader.open(dir); Directory scratch = FSDirectory.open(tmp); IndexWriter writer = new IndexWriter(scratch, conf)) { CodecReader inputs[] = new CodecReader[reader.leaves().size()]; for (int i = 0; i < inputs.length; i++) { inputs[i] = (CodecReader) reader.leaves().get(i).reader(); } writer.addIndexes(inputs); try (DirectoryReader newReader = DirectoryReader.open(writer)) { assert newReader.leaves().size() == 1; SegmentReader sr = (SegmentReader) newReader.leaves().get(0).reader(); report(sr, analyzeFields(sr)); } } finally { IOUtils.rm(tmp); } }
From source file:perf.DiskUsage.java
License:Apache License
public static void main(String args[]) throws Exception { if (args.length != 1) { System.err.println(//from www . j a v a2 s . c om "java [-Djava.io.tmpdir=/scratch] [-Dmode=BEST_COMPRESSION] -cp lucene-core.jar:./build DiskUsage <path to lucene index>"); System.exit(1); } IndexWriterConfig conf = new IndexWriterConfig(null); conf.setOpenMode(OpenMode.CREATE); // force codec to write per-field filenames. conf.setCodec(new Lucene62Codec(Mode.valueOf(System.getProperty("mode", "BEST_SPEED"))) { @Override public PostingsFormat getPostingsFormatForField(String field) { return new Lucene50PostingsFormat(); } @Override public DocValuesFormat getDocValuesFormatForField(String field) { return new Lucene54DocValuesFormat(); } }); Path tmp = Files.createTempDirectory(null); System.err.println("analyzing... (using " + tmp + " for temporary storage)"); try (Directory dir = FSDirectory.open(Paths.get(args[0])); DirectoryReader reader = DirectoryReader.open(dir); Directory scratch = FSDirectory.open(tmp); IndexWriter writer = new IndexWriter(scratch, conf)) { CodecReader inputs[] = new CodecReader[reader.leaves().size()]; for (int i = 0; i < inputs.length; i++) { inputs[i] = (CodecReader) reader.leaves().get(i).reader(); } writer.addIndexes(inputs); try (DirectoryReader newReader = DirectoryReader.open(writer)) { assert newReader.leaves().size() == 1; SegmentReader sr = (SegmentReader) newReader.leaves().get(0).reader(); report(sr, analyzeFields(sr)); } } finally { IOUtils.rm(tmp); } }
From source file:perf.DiskUsage.java
License:Apache License
public static void main(String args[]) throws Exception { if (args.length != 1) { System.err.println(/*from w w w .j a va 2 s . c o m*/ "java [-Djava.io.tmpdir=/scratch] [-Dmode=BEST_COMPRESSION] -cp lucene-core.jar:./build DiskUsage <path to lucene index>"); System.exit(1); } IndexWriterConfig conf = new IndexWriterConfig(null); conf.setOpenMode(OpenMode.CREATE); // force codec to write per-field filenames. conf.setCodec(new Lucene70Codec(Mode.valueOf(System.getProperty("mode", "BEST_SPEED"))) { @Override public PostingsFormat getPostingsFormatForField(String field) { return new Lucene50PostingsFormat(); } @Override public DocValuesFormat getDocValuesFormatForField(String field) { return new Lucene70DocValuesFormat(); } }); Path tmp = Files.createTempDirectory(null); System.err.println("analyzing... (using " + tmp + " for temporary storage)"); try (Directory dir = FSDirectory.open(Paths.get(args[0])); DirectoryReader reader = DirectoryReader.open(dir); Directory scratch = FSDirectory.open(tmp); IndexWriter writer = new IndexWriter(scratch, conf)) { CodecReader inputs[] = new CodecReader[reader.leaves().size()]; for (int i = 0; i < inputs.length; i++) { inputs[i] = (CodecReader) reader.leaves().get(i).reader(); } writer.addIndexes(inputs); try (DirectoryReader newReader = DirectoryReader.open(writer)) { assert newReader.leaves().size() == 1; SegmentReader sr = (SegmentReader) newReader.leaves().get(0).reader(); report(sr, analyzeFields(sr)); } } finally { IOUtils.rm(tmp); } }
From source file:perf.IDPerfTest.java
License:Apache License
private static Result testOne(String indexPath, String desc, IDIterator ids, final int minTermsInBlock, final int maxTermsInBlock) throws IOException { System.out.println("\ntest: " + desc + " termBlocks=" + minTermsInBlock + "/" + maxTermsInBlock); Directory dir = FSDirectory.open(new File(indexPath)); //IndexWriterConfig iwc = new IndexWriterConfig(Version.LUCENE_48, new StandardAnalyzer(Version.LUCENE_48)); IndexWriterConfig iwc = new IndexWriterConfig(Version.LUCENE_4_8, new StandardAnalyzer(Version.LUCENE_4_8)); iwc.setMergeScheduler(new SerialMergeScheduler()); iwc.setOpenMode(IndexWriterConfig.OpenMode.CREATE); // So I can walk the files and get the *.tip sizes: iwc.setUseCompoundFile(false);// w ww .ja v a2 s .com iwc.setCodec(new Lucene53Codec() { @Override public PostingsFormat getPostingsFormatForField(String field) { return new Lucene50PostingsFormat(minTermsInBlock, maxTermsInBlock); } }); /// 7/7/7 segment structure: iwc.setMaxBufferedDocs(ID_COUNT / 777); iwc.setRAMBufferSizeMB(-1); //iwc.setInfoStream(new PrintStreamInfoStream(System.out)); //iwc.setMergePolicy(new LogDocMergePolicy()); ((TieredMergePolicy) iwc.getMergePolicy()).setFloorSegmentMB(.001); ((TieredMergePolicy) iwc.getMergePolicy()).setNoCFSRatio(0.0); //((LogDocMergePolicy) iwc.getMergePolicy()).setMinMergeDocs(1000); iwc.getMergePolicy().setNoCFSRatio(0.0); IndexWriter w = new IndexWriter(dir, iwc); Document doc = new Document(); FieldType ft = new FieldType(StringField.TYPE_NOT_STORED); ft.setTokenized(true); ft.freeze(); BytesRef idValue = new BytesRef(64); Field idField = new Field("id", new BinaryTokenStream(idValue), ft); doc.add(idField); long t0 = System.nanoTime(); BytesRef[] lookupIDs = new BytesRef[ID_SEARCH_COUNT]; Random random = new Random(17); int lookupCount = 0; double rate = 1.01 * ((double) ID_SEARCH_COUNT) / ID_COUNT; for (int i = 0; i < ID_COUNT; i++) { ids.next(idValue); if (lookupCount < lookupIDs.length && random.nextDouble() <= rate) { lookupIDs[lookupCount++] = BytesRef.deepCopyOf(idValue); } // Trickery: the idsIter changed the idValue which the BinaryTokenStream reuses for each added doc w.addDocument(doc); } if (lookupCount < lookupIDs.length) { throw new RuntimeException("didn't get enough lookup ids: " + lookupCount + " vs " + lookupIDs.length); } long indexTime = System.nanoTime() - t0; System.out.println(" indexing done; waitForMerges..."); w.waitForMerges(); IndexReader r = DirectoryReader.open(w, true); System.out.println(" reader=" + r); shuffle(random, lookupIDs); shuffle(random, lookupIDs); long bestTime = Long.MAX_VALUE; long checksum = 0; List<AtomicReaderContext> leaves = new ArrayList<>(r.leaves()); // Sort largest to smallest: Collections.sort(leaves, new Comparator<AtomicReaderContext>() { @Override public int compare(AtomicReaderContext c1, AtomicReaderContext c2) { return c2.reader().maxDoc() - c1.reader().maxDoc(); } }); TermsEnum[] termsEnums = new TermsEnum[leaves.size()]; DocsEnum[] docsEnums = new DocsEnum[leaves.size()]; int[] docBases = new int[leaves.size()]; for (int i = 0; i < leaves.size(); i++) { //System.out.println("i=" + i + " count=" + leaves.get(i).reader().maxDoc()); termsEnums[i] = leaves.get(i).reader().fields().terms("id").iterator(null); docBases[i] = leaves.get(i).docBase; } long rawLookupCount = 0; int countx = 0; for (int iter = 0; iter < 5; iter++) { t0 = System.nanoTime(); BlockTreeTermsReader.seekExactFastNotFound = 0; BlockTreeTermsReader.seekExactFastRootNotFound = 0; rawLookupCount = 0; for (BytesRef id : lookupIDs) { if (countx++ < 50) { System.out.println(" id=" + id); } boolean found = false; for (int seg = 0; seg < termsEnums.length; seg++) { rawLookupCount++; if (termsEnums[seg].seekExact(id)) { docsEnums[seg] = termsEnums[seg].docs(null, docsEnums[seg], 0); int docID = docsEnums[seg].nextDoc(); if (docID == DocsEnum.NO_MORE_DOCS) { // uh-oh! throw new RuntimeException("id not found: " + id); } // paranoia: checksum += docID + docBases[seg]; found = true; // Optimization vs MultiFields: we don't need to check any more segments since id is PK break; } } if (found == false) { // uh-oh! throw new RuntimeException("id not found: " + id); } } long lookupTime = System.nanoTime() - t0; System.out.println(String.format(Locale.ROOT, " iter=" + iter + " lookupTime=%.3f sec", lookupTime / 1000000000.0)); if (lookupTime < bestTime) { bestTime = lookupTime; System.out.println(" **"); } } long totalBytes = 0; long termsIndexTotalBytes = 0; for (String fileName : dir.listAll()) { long bytes = dir.fileLength(fileName); totalBytes += bytes; if (fileName.endsWith(".tip")) { termsIndexTotalBytes += bytes; } } r.close(); w.rollback(); dir.close(); return new Result(desc, ID_COUNT / (indexTime / 1000000.0), lookupIDs.length / (bestTime / 1000000.0), totalBytes, termsIndexTotalBytes, checksum, BlockTreeTermsReader.seekExactFastNotFound, BlockTreeTermsReader.seekExactFastRootNotFound, rawLookupCount, minTermsInBlock, maxTermsInBlock); }
From source file:perf.IndexAndSearchOpenStreetMaps.java
License:Apache License
private static void createIndex(boolean fast, boolean doForceMerge, boolean doDistanceSort) throws IOException, InterruptedException { CharsetDecoder decoder = StandardCharsets.UTF_8.newDecoder().onMalformedInput(CodingErrorAction.REPORT) .onUnmappableCharacter(CodingErrorAction.REPORT); int BUFFER_SIZE = 1 << 16; // 64K InputStream is;/* w ww . j a v a2 s . co m*/ if (SMALL) { is = Files.newInputStream(Paths.get(DATA_LOCATION, "latlon.subsetPlusAllLondon.txt")); } else { is = Files.newInputStream(Paths.get(DATA_LOCATION, "latlon.txt")); } BufferedReader reader = new BufferedReader(new InputStreamReader(is, decoder), BUFFER_SIZE); int NUM_THREADS; if (fast) { NUM_THREADS = 4; } else { NUM_THREADS = 1; } int CHUNK = 10000; long t0 = System.nanoTime(); AtomicLong totalCount = new AtomicLong(); for (int part = 0; part < NUM_PARTS; part++) { Directory dir = FSDirectory.open(Paths.get(getName(part, doDistanceSort))); IndexWriterConfig iwc = new IndexWriterConfig(null); iwc.setCodec(getCodec(fast)); iwc.setOpenMode(IndexWriterConfig.OpenMode.CREATE); if (fast) { ((TieredMergePolicy) iwc.getMergePolicy()).setMaxMergedSegmentMB(Double.POSITIVE_INFINITY); iwc.setRAMBufferSizeMB(1024); } else { iwc.setMaxBufferedDocs(109630); iwc.setMergePolicy(new LogDocMergePolicy()); iwc.setMergeScheduler(new SerialMergeScheduler()); } iwc.setInfoStream(new PrintStreamInfoStream(System.out)); IndexWriter w = new IndexWriter(dir, iwc); Thread[] threads = new Thread[NUM_THREADS]; AtomicBoolean finished = new AtomicBoolean(); Object lock = new Object(); final int finalPart = part; for (int t = 0; t < NUM_THREADS; t++) { threads[t] = new Thread() { @Override public void run() { String[] lines = new String[CHUNK]; int chunkCount = 0; while (finished.get() == false) { try { int count = CHUNK; synchronized (lock) { for (int i = 0; i < CHUNK; i++) { String line = reader.readLine(); if (line == null) { count = i; finished.set(true); break; } lines[i] = line; } if (finalPart == 0 && totalCount.get() + count >= 2000000000) { finished.set(true); } } for (int i = 0; i < count; i++) { String[] parts = lines[i].split(","); //long id = Long.parseLong(parts[0]); double lat = Double.parseDouble(parts[1]); double lon = Double.parseDouble(parts[2]); Document doc = new Document(); if (useGeoPoint) { doc.add(new GeoPointField("point", lat, lon, Field.Store.NO)); } else if (useGeo3D || useGeo3DLarge) { doc.add(new Geo3DPoint("point", lat, lon)); } else { doc.add(new LatLonPoint("point", lat, lon)); if (doDistanceSort) { doc.add(new LatLonDocValuesField("point", lat, lon)); } } w.addDocument(doc); long x = totalCount.incrementAndGet(); if (x % 1000000 == 0) { System.out.println(x + "..."); } } chunkCount++; if (false && SMALL == false && chunkCount == 20000) { System.out.println("NOW BREAK EARLY"); break; } } catch (IOException ioe) { throw new RuntimeException(ioe); } } } }; threads[t].start(); } for (Thread thread : threads) { thread.join(); } System.out.println("Part " + part + " is done: w.maxDoc()=" + w.maxDoc()); w.commit(); System.out.println("done commit"); long t1 = System.nanoTime(); System.out.println(((t1 - t0) / 1000000000.0) + " sec to index part " + part); if (doForceMerge) { w.forceMerge(1); long t2 = System.nanoTime(); System.out.println(((t2 - t1) / 1000000000.0) + " sec to force merge part " + part); } w.close(); } //System.out.println(totalCount.get() + " total docs"); //System.out.println("Force merge..."); //w.forceMerge(1); //long t2 = System.nanoTime(); //System.out.println(((t2-t1)/1000000000.0) + " sec to force merge"); //w.close(); //long t3 = System.nanoTime(); //System.out.println(((t3-t2)/1000000000.0) + " sec to close"); //System.out.println(((t3-t2)/1000000000.0) + " sec to close"); }
From source file:perf.IndexAndSearchOpenStreetMapsGeo3D.java
License:Apache License
private static void createIndex() throws IOException { long t0 = System.nanoTime(); CharsetDecoder decoder = StandardCharsets.UTF_8.newDecoder().onMalformedInput(CodingErrorAction.REPORT) .onUnmappableCharacter(CodingErrorAction.REPORT); int BUFFER_SIZE = 1 << 16; // 64K InputStream is = Files .newInputStream(Paths.get("/lucenedata/open-street-maps/latlon.subsetPlusAllLondon.txt")); BufferedReader reader = new BufferedReader(new InputStreamReader(is, decoder), BUFFER_SIZE); DocValuesFormat dvFormat = new Geo3DDocValuesFormat(); Directory dir = FSDirectory.open(Paths.get("bkdtestgeo3d")); Codec codec = new Lucene53Codec() { @Override/* w w w .j av a 2 s . c om*/ public DocValuesFormat getDocValuesFormatForField(String field) { return dvFormat; } }; IndexWriterConfig iwc = new IndexWriterConfig(null); //iwc.setRAMBufferSizeMB(256); iwc.setMaxBufferedDocs(109630); iwc.setCodec(codec); iwc.setOpenMode(IndexWriterConfig.OpenMode.CREATE); IndexWriter w = new IndexWriter(dir, iwc); while (true) { String line = reader.readLine(); if (line == null) { break; } String[] parts = line.split(","); //long id = Long.parseLong(parts[0]); double lat = toRadians(Double.parseDouble(parts[1])); double lon = toRadians(Double.parseDouble(parts[2])); Document doc = new Document(); doc.add(new Geo3DPointField("point", PlanetModel.WGS84, lat, lon)); w.addDocument(doc); } long t1 = System.nanoTime(); System.out.println(((t1 - t0) / 1000000000.0) + " sec to build index"); System.out.println(w.maxDoc() + " total docs"); //w.forceMerge(1); w.close(); long t2 = System.nanoTime(); System.out.println(((t2 - t1) / 1000000000.0) + " sec to close"); }
From source file:perf.Indexer.java
License:Apache License
private static void _main(String[] clArgs) throws Exception { Args args = new Args(clArgs); // EG: -facets Date -facets characterCount ... FacetsConfig facetsConfig = new FacetsConfig(); facetsConfig.setHierarchical("Date", true); final Set<String> facetFields = new HashSet<String>(); if (args.hasArg("-facets")) { for (String arg : args.getStrings("-facets")) { facetFields.add(arg);//w w w . ja v a2 s. com } } final String dirImpl = args.getString("-dirImpl"); final String dirPath = args.getString("-indexPath") + "/index"; final Directory dir; OpenDirectory od = OpenDirectory.get(dirImpl); dir = od.open(Paths.get(dirPath)); final String analyzer = args.getString("-analyzer"); final Analyzer a; if (analyzer.equals("EnglishAnalyzer")) { a = new EnglishAnalyzer(); } else if (analyzer.equals("StandardAnalyzer")) { a = new StandardAnalyzer(); } else if (analyzer.equals("StandardAnalyzerNoStopWords")) { a = new StandardAnalyzer(CharArraySet.EMPTY_SET); } else if (analyzer.equals("ShingleStandardAnalyzer")) { a = new ShingleAnalyzerWrapper(new StandardAnalyzer(), 2, 2); } else if (analyzer.equals("ShingleStandardAnalyzerNoStopWords")) { a = new ShingleAnalyzerWrapper(new StandardAnalyzer(CharArraySet.EMPTY_SET), 2, 2); } else { throw new RuntimeException("unknown analyzer " + analyzer); } final String lineFile = args.getString("-lineDocsFile"); // -1 means all docs in the line file: final int docCountLimit = args.getInt("-docCountLimit"); final int numThreads = args.getInt("-threadCount"); final boolean doForceMerge = args.getFlag("-forceMerge"); final boolean verbose = args.getFlag("-verbose"); String indexSortField = null; SortField.Type indexSortType = null; if (args.hasArg("-indexSort")) { indexSortField = args.getString("-indexSort"); int i = indexSortField.indexOf(':'); if (i == -1) { throw new IllegalArgumentException( "-indexSort should have form field:type; got: " + indexSortField); } String typeString = indexSortField.substring(i + 1, indexSortField.length()); if (typeString.equals("long")) { indexSortType = SortField.Type.LONG; } else if (typeString.equals("string")) { indexSortType = SortField.Type.STRING; } else { throw new IllegalArgumentException("-indexSort can only handle 'long' sort; got: " + typeString); } indexSortField = indexSortField.substring(0, i); } else { indexSortType = null; } final double ramBufferSizeMB = args.getDouble("-ramBufferMB"); final int maxBufferedDocs = args.getInt("-maxBufferedDocs"); final String defaultPostingsFormat = args.getString("-postingsFormat"); final boolean doDeletions = args.getFlag("-deletions"); final boolean printDPS = args.getFlag("-printDPS"); final boolean waitForMerges = args.getFlag("-waitForMerges"); final boolean waitForCommit = args.getFlag("-waitForCommit"); final String mergePolicy = args.getString("-mergePolicy"); final Mode mode; final boolean doUpdate = args.getFlag("-update"); if (doUpdate) { mode = Mode.UPDATE; } else { mode = Mode.valueOf(args.getString("-mode", "add").toUpperCase(Locale.ROOT)); } int randomDocIDMax; if (mode == Mode.UPDATE) { randomDocIDMax = args.getInt("-randomDocIDMax"); } else { randomDocIDMax = -1; } final String idFieldPostingsFormat = args.getString("-idFieldPostingsFormat"); final boolean addGroupingFields = args.getFlag("-grouping"); final boolean useCFS = args.getFlag("-cfs"); final boolean storeBody = args.getFlag("-store"); final boolean tvsBody = args.getFlag("-tvs"); final boolean bodyPostingsOffsets = args.getFlag("-bodyPostingsOffsets"); final int maxConcurrentMerges = args.getInt("-maxConcurrentMerges"); final boolean addDVFields = args.getFlag("-dvfields"); final boolean doRandomCommit = args.getFlag("-randomCommit"); final boolean useCMS = args.getFlag("-useCMS"); final boolean disableIOThrottle = args.getFlag("-disableIOThrottle"); if (waitForCommit == false && waitForMerges) { throw new RuntimeException("pass -waitForCommit if you pass -waitForMerges"); } if (waitForCommit == false && doForceMerge) { throw new RuntimeException("pass -waitForCommit if you pass -forceMerge"); } if (waitForCommit == false && doDeletions) { throw new RuntimeException("pass -waitForCommit if you pass -deletions"); } if (useCMS == false && disableIOThrottle) { throw new RuntimeException("-disableIOThrottle only makes sense with -useCMS"); } final double nrtEverySec; if (args.hasArg("-nrtEverySec")) { nrtEverySec = args.getDouble("-nrtEverySec"); } else { nrtEverySec = -1.0; } // True to start back at the beginning if we run out of // docs from the line file source: final boolean repeatDocs = args.getFlag("-repeatDocs"); final String facetDVFormatName; if (facetFields.isEmpty()) { facetDVFormatName = "Lucene54"; } else { facetDVFormatName = args.getString("-facetDVFormat"); } if (addGroupingFields && docCountLimit == -1) { a.close(); throw new RuntimeException("cannot add grouping fields unless docCount is set"); } args.check(); System.out.println("Dir: " + dirImpl); System.out.println("Index path: " + dirPath); System.out.println("Analyzer: " + analyzer); System.out.println("Line file: " + lineFile); System.out.println("Doc count limit: " + (docCountLimit == -1 ? "all docs" : "" + docCountLimit)); System.out.println("Threads: " + numThreads); System.out.println("Force merge: " + (doForceMerge ? "yes" : "no")); System.out.println("Verbose: " + (verbose ? "yes" : "no")); System.out.println("RAM Buffer MB: " + ramBufferSizeMB); System.out.println("Max buffered docs: " + maxBufferedDocs); System.out.println("Default postings format: " + defaultPostingsFormat); System.out.println("Do deletions: " + (doDeletions ? "yes" : "no")); System.out.println("Wait for merges: " + (waitForMerges ? "yes" : "no")); System.out.println("Wait for commit: " + (waitForCommit ? "yes" : "no")); System.out.println("IO throttle: " + (disableIOThrottle ? "no" : "yes")); System.out.println("Merge policy: " + mergePolicy); System.out.println("Mode: " + mode); if (mode == Mode.UPDATE) { System.out.println("DocIDMax: " + randomDocIDMax); } System.out.println("ID field postings format: " + idFieldPostingsFormat); System.out.println("Add grouping fields: " + (addGroupingFields ? "yes" : "no")); System.out.println("Compound file format: " + (useCFS ? "yes" : "no")); System.out.println("Store body field: " + (storeBody ? "yes" : "no")); System.out.println("Term vectors for body field: " + (tvsBody ? "yes" : "no")); System.out.println("Facet DV Format: " + facetDVFormatName); System.out.println("Facet fields: " + facetFields); System.out.println("Body postings offsets: " + (bodyPostingsOffsets ? "yes" : "no")); System.out.println("Max concurrent merges: " + maxConcurrentMerges); System.out.println("Add DocValues fields: " + addDVFields); System.out.println("Use ConcurrentMergeScheduler: " + useCMS); if (nrtEverySec > 0.0) { System.out.println("Open & close NRT reader every: " + nrtEverySec + " sec"); } else { System.out.println("Open & close NRT reader every: never"); } System.out.println("Repeat docs: " + repeatDocs); if (verbose) { InfoStream.setDefault(new PrintStreamInfoStream(System.out)); } final IndexWriterConfig iwc = new IndexWriterConfig(a); if (indexSortField != null) { iwc.setIndexSort(new Sort(new SortField(indexSortField, indexSortType))); } if (mode == Mode.UPDATE) { iwc.setOpenMode(IndexWriterConfig.OpenMode.CREATE_OR_APPEND); } else { iwc.setOpenMode(IndexWriterConfig.OpenMode.CREATE); } iwc.setMaxBufferedDocs(maxBufferedDocs); iwc.setRAMBufferSizeMB(ramBufferSizeMB); // So flushed segments do/don't use CFS: iwc.setUseCompoundFile(useCFS); final AtomicBoolean indexingFailed = new AtomicBoolean(); iwc.setMergeScheduler(getMergeScheduler(indexingFailed, useCMS, maxConcurrentMerges, disableIOThrottle)); iwc.setMergePolicy(getMergePolicy(mergePolicy, useCFS)); // Keep all commit points: if (doDeletions || doForceMerge) { iwc.setIndexDeletionPolicy(NoDeletionPolicy.INSTANCE); } final Codec codec = new Lucene62Codec() { @Override public PostingsFormat getPostingsFormatForField(String field) { return PostingsFormat.forName(field.equals("id") ? idFieldPostingsFormat : defaultPostingsFormat); } private final DocValuesFormat facetsDVFormat = DocValuesFormat.forName(facetDVFormatName); //private final DocValuesFormat lucene42DVFormat = DocValuesFormat.forName("Lucene42"); //private final DocValuesFormat diskDVFormat = DocValuesFormat.forName("Disk"); // private final DocValuesFormat lucene45DVFormat = DocValuesFormat.forName("Lucene45"); private final DocValuesFormat directDVFormat = DocValuesFormat.forName("Direct"); @Override public DocValuesFormat getDocValuesFormatForField(String field) { if (facetFields.contains(field) || field.equals("$facets")) { return facetsDVFormat; //} else if (field.equals("$facets_sorted_doc_values")) { //return diskDVFormat; } else { // Use default DVFormat for all else: // System.out.println("DV: field=" + field + " format=" + super.getDocValuesFormatForField(field)); return super.getDocValuesFormatForField(field); } } }; iwc.setCodec(codec); System.out.println("IW config=" + iwc); IndexWriter w = new IndexWriter(dir, iwc); System.out.println("Index has " + w.maxDoc() + " docs"); final TaxonomyWriter taxoWriter; if (facetFields.isEmpty() == false) { taxoWriter = new DirectoryTaxonomyWriter(od.open(Paths.get(args.getString("-indexPath"), "facets")), IndexWriterConfig.OpenMode.CREATE); } else { taxoWriter = null; } // Fixed seed so group field values are always consistent: final Random random = new Random(17); LineFileDocs lineFileDocs = new LineFileDocs(lineFile, repeatDocs, storeBody, tvsBody, bodyPostingsOffsets, false, taxoWriter, facetFields, facetsConfig, addDVFields); float docsPerSecPerThread = -1f; //float docsPerSecPerThread = 100f; IndexThreads threads = new IndexThreads(random, w, indexingFailed, lineFileDocs, numThreads, docCountLimit, addGroupingFields, printDPS, mode, docsPerSecPerThread, null, nrtEverySec, randomDocIDMax); System.out.println("\nIndexer: start"); final long t0 = System.currentTimeMillis(); threads.start(); while (!threads.done() && indexingFailed.get() == false) { Thread.sleep(100); // Commits once per minute on average: if (doRandomCommit && random.nextInt(600) == 17) { System.out.println("Indexer: now commit"); long commitStartNS = System.nanoTime(); w.commit(); System.out.println(String.format(Locale.ROOT, "Indexer: commit took %.1f msec", (System.nanoTime() - commitStartNS) / 1000000.)); } } threads.stop(); final long t1 = System.currentTimeMillis(); System.out.println("\nIndexer: indexing done (" + (t1 - t0) + " msec); total " + w.maxDoc() + " docs"); // if we update we can not tell how many docs if (threads.failed.get()) { throw new RuntimeException("exceptions during indexing"); } if (mode != Mode.UPDATE && docCountLimit != -1 && w.maxDoc() != docCountLimit) { throw new RuntimeException("w.maxDoc()=" + w.maxDoc() + " but expected " + docCountLimit); } final Map<String, String> commitData = new HashMap<String, String>(); if (waitForMerges) { w.close(); IndexWriterConfig iwc2 = new IndexWriterConfig(a); iwc2.setMergeScheduler( getMergeScheduler(indexingFailed, useCMS, maxConcurrentMerges, disableIOThrottle)); iwc2.setMergePolicy(getMergePolicy(mergePolicy, useCFS)); iwc2.setCodec(codec); iwc2.setUseCompoundFile(useCFS); iwc2.setMaxBufferedDocs(maxBufferedDocs); iwc2.setRAMBufferSizeMB(ramBufferSizeMB); if (indexSortField != null) { iwc2.setIndexSort(new Sort(new SortField(indexSortField, indexSortType))); } w = new IndexWriter(dir, iwc2); long t2 = System.currentTimeMillis(); System.out.println("\nIndexer: waitForMerges done (" + (t2 - t1) + " msec)"); } if (waitForCommit) { commitData.put("userData", "multi"); w.setLiveCommitData(commitData.entrySet()); long t2 = System.currentTimeMillis(); w.commit(); long t3 = System.currentTimeMillis(); System.out.println("\nIndexer: commit multi (took " + (t3 - t2) + " msec)"); } else { w.rollback(); w = null; } if (doForceMerge) { long forceMergeStartMSec = System.currentTimeMillis(); w.forceMerge(1); long forceMergeEndMSec = System.currentTimeMillis(); System.out.println( "\nIndexer: force merge done (took " + (forceMergeEndMSec - forceMergeStartMSec) + " msec)"); commitData.put("userData", "single"); w.setLiveCommitData(commitData.entrySet()); w.commit(); final long t5 = System.currentTimeMillis(); System.out.println("\nIndexer: commit single done (took " + (t5 - forceMergeEndMSec) + " msec)"); } if (doDeletions) { final long t5 = System.currentTimeMillis(); // Randomly delete 5% of the docs final Set<Integer> deleted = new HashSet<Integer>(); final int maxDoc = w.maxDoc(); final int toDeleteCount = (int) (maxDoc * 0.05); System.out.println("\nIndexer: delete " + toDeleteCount + " docs"); while (deleted.size() < toDeleteCount) { final int id = random.nextInt(maxDoc); if (!deleted.contains(id)) { deleted.add(id); w.deleteDocuments(new Term("id", LineFileDocs.intToID(id))); } } final long t6 = System.currentTimeMillis(); System.out.println("\nIndexer: deletes done (took " + (t6 - t5) + " msec)"); commitData.put("userData", doForceMerge ? "delsingle" : "delmulti"); w.setLiveCommitData(commitData.entrySet()); w.commit(); final long t7 = System.currentTimeMillis(); System.out.println("\nIndexer: commit delmulti done (took " + (t7 - t6) + " msec)"); if (doUpdate || w.numDocs() != maxDoc - toDeleteCount) { throw new RuntimeException( "count mismatch: w.numDocs()=" + w.numDocs() + " but expected " + (maxDoc - toDeleteCount)); } } if (taxoWriter != null) { System.out.println("Taxonomy has " + taxoWriter.getSize() + " ords"); taxoWriter.commit(); taxoWriter.close(); } final long tCloseStart = System.currentTimeMillis(); if (w != null) { w.close(); w = null; } if (waitForCommit) { System.out.println("\nIndexer: at close: " + SegmentInfos.readLatestCommit(dir)); System.out.println("\nIndexer: close took " + (System.currentTimeMillis() - tCloseStart) + " msec"); } dir.close(); final long tFinal = System.currentTimeMillis(); System.out.println("\nIndexer: net bytes indexed " + threads.getBytesIndexed()); final long indexingTime; if (waitForCommit) { indexingTime = tFinal - t0; System.out.println("\nIndexer: finished (" + indexingTime + " msec)"); } else { indexingTime = t1 - t0; System.out.println("\nIndexer: finished (" + indexingTime + " msec), excluding commit"); } System.out.println( "\nIndexer: " + (threads.getBytesIndexed() / 1024. / 1024. / 1024. / (indexingTime / 3600000.)) + " GB/hour plain text"); }
From source file:perf.NRTPerfTest.java
License:Apache License
public static void main(String[] args) throws Exception { final String dirImpl = args[0]; final String dirPath = args[1]; final String commit = args[2]; final String lineDocFile = args[3]; final long seed = Long.parseLong(args[4]); final double docsPerSec = Double.parseDouble(args[5]); final double runTimeSec = Double.parseDouble(args[6]); final int numSearchThreads = Integer.parseInt(args[7]); int numIndexThreads = Integer.parseInt(args[8]); if (numIndexThreads > docsPerSec) { System.out.println("INFO: numIndexThreads higher than docsPerSec, adjusting numIndexThreads"); numIndexThreads = (int) Math.max(1, docsPerSec); }/*from w ww . jav a 2 s .com*/ final double reopenPerSec = Double.parseDouble(args[9]); final Mode mode = Mode.valueOf(args[10].toUpperCase(Locale.ROOT)); statsEverySec = Integer.parseInt(args[11]); final boolean doCommit = args[12].equals("yes"); final double mergeMaxWriteMBPerSec = Double.parseDouble(args[13]); if (mergeMaxWriteMBPerSec != 0.0) { throw new IllegalArgumentException("mergeMaxWriteMBPerSec must be 0.0 until LUCENE-3202 is done"); } final String tasksFile = args[14]; if (Files.notExists(Paths.get(tasksFile))) { throw new FileNotFoundException("tasks file not found " + tasksFile); } final boolean hasProcMemInfo = Files.exists(Paths.get("/proc/meminfo")); System.out.println("DIR=" + dirImpl); System.out.println("Index=" + dirPath); System.out.println("Commit=" + commit); System.out.println("LineDocs=" + lineDocFile); System.out.println("Docs/sec=" + docsPerSec); System.out.println("Run time sec=" + runTimeSec); System.out.println("NumSearchThreads=" + numSearchThreads); System.out.println("NumIndexThreads=" + numIndexThreads); System.out.println("Reopen/sec=" + reopenPerSec); System.out.println("Mode=" + mode); System.out.println("tasksFile=" + tasksFile); System.out.println("Record stats every " + statsEverySec + " seconds"); final int count = (int) ((runTimeSec / statsEverySec) + 2); docsIndexedByTime = new AtomicInteger[count]; searchesByTime = new AtomicInteger[count]; totalUpdateTimeByTime = new AtomicLong[count]; final AtomicInteger reopensByTime[] = new AtomicInteger[count]; for (int i = 0; i < count; i++) { docsIndexedByTime[i] = new AtomicInteger(); searchesByTime[i] = new AtomicInteger(); totalUpdateTimeByTime[i] = new AtomicLong(); reopensByTime[i] = new AtomicInteger(); } System.out.println( "Max merge MB/sec = " + (mergeMaxWriteMBPerSec <= 0.0 ? "unlimited" : mergeMaxWriteMBPerSec)); final Random random = new Random(seed); final LineFileDocs docs = new LineFileDocs(lineDocFile, true, false, false, false, false, null, new HashSet<String>(), null, true); final Directory dir0; if (dirImpl.equals("MMapDirectory")) { dir0 = new MMapDirectory(Paths.get(dirPath)); } else if (dirImpl.equals("NIOFSDirectory")) { dir0 = new NIOFSDirectory(Paths.get(dirPath)); } else if (dirImpl.equals("SimpleFSDirectory")) { dir0 = new SimpleFSDirectory(Paths.get(dirPath)); } else { docs.close(); throw new RuntimeException("unknown directory impl \"" + dirImpl + "\""); } //final NRTCachingDirectory dir = new NRTCachingDirectory(dir0, 10, 200.0, mergeMaxWriteMBPerSec); final NRTCachingDirectory dir = new NRTCachingDirectory(dir0, 20, 400.0); //final MergeScheduler ms = dir.getMergeScheduler(); //final Directory dir = dir0; //final MergeScheduler ms = new ConcurrentMergeScheduler(); final String field = "body"; // Open an IW on the requested commit point, but, don't // delete other (past or future) commit points: // TODO take Analyzer as parameter StandardAnalyzer analyzer = new StandardAnalyzer(CharArraySet.EMPTY_SET); final IndexWriterConfig conf = new IndexWriterConfig(analyzer); conf.setIndexDeletionPolicy(NoDeletionPolicy.INSTANCE); conf.setRAMBufferSizeMB(256.0); //iwc.setMergeScheduler(ms); final Codec codec = new Lucene62Codec() { @Override public PostingsFormat getPostingsFormatForField(String field) { if (field.equals("id")) { return PostingsFormat.forName("Memory"); } else { return PostingsFormat.forName("Lucene50"); } } private final DocValuesFormat direct = DocValuesFormat.forName("Direct"); @Override public DocValuesFormat getDocValuesFormatForField(String field) { return direct; } }; conf.setCodec(codec); /* iwc.setMergePolicy(new LogByteSizeMergePolicy()); ((LogMergePolicy) iwc.getMergePolicy()).setUseCompoundFile(false); ((LogMergePolicy) iwc.getMergePolicy()).setMergeFactor(30); ((LogByteSizeMergePolicy) iwc.getMergePolicy()).setMaxMergeMB(10000.0); System.out.println("USING LOG BS MP"); */ TieredMergePolicy tmp = new TieredMergePolicy(); tmp.setNoCFSRatio(0.0); tmp.setMaxMergedSegmentMB(1000000.0); //tmp.setReclaimDeletesWeight(3.0); //tmp.setMaxMergedSegmentMB(7000.0); conf.setMergePolicy(tmp); if (!commit.equals("none")) { conf.setIndexCommit(PerfUtils.findCommitPoint(commit, dir)); } // Make sure merges run @ higher prio than indexing: final ConcurrentMergeScheduler cms = (ConcurrentMergeScheduler) conf.getMergeScheduler(); cms.setMaxMergesAndThreads(4, 1); conf.setMergedSegmentWarmer(new MergedReaderWarmer(field)); final IndexWriter w = new IndexWriter(dir, conf); // w.setInfoStream(System.out); IndexThreads.UpdatesListener updatesListener = new IndexThreads.UpdatesListener() { long startTimeNS; @Override public void beforeUpdate() { startTimeNS = System.nanoTime(); } @Override public void afterUpdate() { int idx = currentQT.get(); totalUpdateTimeByTime[idx].addAndGet(System.nanoTime() - startTimeNS); docsIndexedByTime[idx].incrementAndGet(); } }; IndexThreads indexThreads = new IndexThreads(random, w, new AtomicBoolean(false), docs, numIndexThreads, -1, false, false, mode, (float) (docsPerSec / numIndexThreads), updatesListener, -1.0, w.maxDoc()); // NativePosixUtil.mlockTermsDict(startR, "id"); final SearcherManager manager = new SearcherManager(w, null); IndexSearcher s = manager.acquire(); try { System.out.println("Reader=" + s.getIndexReader()); } finally { manager.release(s); } final DirectSpellChecker spellChecker = new DirectSpellChecker(); final IndexState indexState = new IndexState(manager, null, field, spellChecker, "PostingsHighlighter", null); final QueryParser qp = new QueryParser(field, analyzer); TaskParser taskParser = new TaskParser(indexState, qp, field, 10, random, true); final TaskSource tasks = new RandomTaskSource(taskParser, tasksFile, random) { @Override public void taskDone(Task task, long queueTimeNS, int toalHitCount) { searchesByTime[currentQT.get()].incrementAndGet(); } }; System.out.println("Task repeat count 1"); System.out.println("Tasks file " + tasksFile); System.out.println("Num task per cat 20"); final TaskThreads taskThreads = new TaskThreads(tasks, indexState, numSearchThreads); final ReopenThread reopenThread = new ReopenThread(reopenPerSec, manager, reopensByTime, runTimeSec); reopenThread.setName("ReopenThread"); reopenThread.setPriority(4 + Thread.currentThread().getPriority()); System.out.println("REOPEN PRI " + reopenThread.getPriority()); indexThreads.start(); reopenThread.start(); taskThreads.start(); Thread.currentThread().setPriority(5 + Thread.currentThread().getPriority()); System.out.println("TIMER PRI " + Thread.currentThread().getPriority()); //System.out.println("Start: " + new Date()); final long startMS = System.currentTimeMillis(); final long stopMS = startMS + (long) (runTimeSec * 1000); int lastQT = -1; while (true) { final long t = System.currentTimeMillis(); if (t >= stopMS) { break; } final int qt = (int) ((t - startMS) / statsEverySec / 1000); currentQT.set(qt); if (qt != lastQT) { final int prevQT = lastQT; lastQT = qt; if (prevQT > 0) { final String other; if (hasProcMemInfo) { other = " D=" + getLinuxDirtyBytes(); } else { other = ""; } int prev = prevQT - 1; System.out.println(String.format("QT %d searches=%d docs=%d reopens=%s totUpdateTime=%d", prev, searchesByTime[prev].get(), docsIndexedByTime[prev].get(), reopensByTime[prev].get() + other, TimeUnit.NANOSECONDS.toMillis(totalUpdateTimeByTime[prev].get()))); } } Thread.sleep(25); } taskThreads.stop(); reopenThread.join(); indexThreads.stop(); System.out.println("By time:"); for (int i = 0; i < searchesByTime.length - 2; i++) { System.out.println(String.format(" %d searches=%d docs=%d reopens=%d totUpdateTime=%d", i * statsEverySec, searchesByTime[i].get(), docsIndexedByTime[i].get(), reopensByTime[i].get(), TimeUnit.NANOSECONDS.toMillis(totalUpdateTimeByTime[i].get()))); } manager.close(); if (doCommit) { w.close(); } else { w.rollback(); } }
From source file:perf.SearchPerfTest.java
License:Apache License
private static void _main(String[] clArgs) throws Exception { // args: dirImpl indexPath numThread numIterPerThread // eg java SearchPerfTest /path/to/index 4 100 final Args args = new Args(clArgs); Directory dir0;/* w w w .ja v a 2s. c o m*/ final String dirPath = args.getString("-indexPath") + "/index"; final String dirImpl = args.getString("-dirImpl"); OpenDirectory od = OpenDirectory.get(dirImpl); /* } else if (dirImpl.equals("NativePosixMMapDirectory")) { dir0 = new NativePosixMMapDirectory(new File(dirPath)); ramDir = null; if (doFacets) { facetsDir = new NativePosixMMapDirectory(new File(facetsDirPath)); } } else if (dirImpl.equals("CachingDirWrapper")) { dir0 = new CachingRAMDirectory(new MMapDirectory(new File(dirPath))); ramDir = null; } else if (dirImpl.equals("RAMExceptDirectPostingsDirectory")) { // Load only non-postings files into RAMDir (assumes // Lucene40PF is the wrapped PF): Set<String> postingsExtensions = new HashSet<String>(); postingsExtensions.add("frq"); postingsExtensions.add("prx"); postingsExtensions.add("tip"); postingsExtensions.add("tim"); ramDir = new RAMDirectory(); Directory fsDir = new MMapDirectory(new File(dirPath)); for (String file : fsDir.listAll()) { int idx = file.indexOf('.'); if (idx != -1 && postingsExtensions.contains(file.substring(idx+1, file.length()))) { continue; } fsDir.copy(ramDir, file, file, IOContext.READ); } dir0 = new FileSwitchDirectory(postingsExtensions, fsDir, ramDir, true); if (doFacets) { facetsDir = new RAMDirectory(new SimpleFSDirectory(new File(facetsDirPath)), IOContext.READ); } */ final RAMDirectory ramDir; dir0 = od.open(Paths.get(dirPath)); if (dir0 instanceof RAMDirectory) { ramDir = (RAMDirectory) dir0; } else { ramDir = null; } // TODO: NativeUnixDir? final String analyzer = args.getString("-analyzer"); final String tasksFile = args.getString("-taskSource"); final int searchThreadCount = args.getInt("-searchThreadCount"); final String fieldName = args.getString("-field"); final boolean printHeap = args.getFlag("-printHeap"); final boolean doPKLookup = args.getFlag("-pk"); final int topN = args.getInt("-topN"); final boolean doStoredLoads = args.getFlag("-loadStoredFields"); // Used to choose which random subset of tasks we will // run, to generate the PKLookup tasks, and to generate // any random pct filters: final long staticRandomSeed = args.getLong("-staticSeed"); // Used to shuffle the random subset of tasks: final long randomSeed = args.getLong("-seed"); // TODO: this could be way better. final String similarity = args.getString("-similarity"); // now reflect final Class<? extends Similarity> simClazz = Class .forName("org.apache.lucene.search.similarities." + similarity).asSubclass(Similarity.class); final Similarity sim = simClazz.newInstance(); System.out.println("Using dir impl " + dir0.getClass().getName()); System.out.println("Analyzer " + analyzer); System.out.println("Similarity " + similarity); System.out.println("Search thread count " + searchThreadCount); System.out.println("topN " + topN); System.out.println("JVM " + (Constants.JRE_IS_64BIT ? "is" : "is not") + " 64bit"); System.out.println("Pointer is " + RamUsageEstimator.NUM_BYTES_OBJECT_REF + " bytes"); final Analyzer a; if (analyzer.equals("EnglishAnalyzer")) { a = new EnglishAnalyzer(); } else if (analyzer.equals("ClassicAnalyzer")) { a = new ClassicAnalyzer(); } else if (analyzer.equals("StandardAnalyzer")) { a = new StandardAnalyzer(); } else if (analyzer.equals("StandardAnalyzerNoStopWords")) { a = new StandardAnalyzer(CharArraySet.EMPTY_SET); } else if (analyzer.equals("ShingleStandardAnalyzer")) { a = new ShingleAnalyzerWrapper(new StandardAnalyzer(CharArraySet.EMPTY_SET), 2, 2, ShingleFilter.DEFAULT_TOKEN_SEPARATOR, true, true, ShingleFilter.DEFAULT_FILLER_TOKEN); } else { throw new RuntimeException("unknown analyzer " + analyzer); } final ReferenceManager<IndexSearcher> mgr; final IndexWriter writer; final Directory dir; final String commit = args.getString("-commit"); final String hiliteImpl = args.getString("-hiliteImpl"); final String logFile = args.getString("-log"); final long tSearcherStart = System.currentTimeMillis(); final boolean verifyCheckSum = !args.getFlag("-skipVerifyChecksum"); final boolean recacheFilterDeletes = args.getFlag("-recacheFilterDeletes"); if (recacheFilterDeletes) { throw new UnsupportedOperationException("recacheFilterDeletes was deprecated"); } if (args.getFlag("-nrt")) { // TODO: get taxoReader working here too // TODO: factor out & share this CL processing w/ Indexer final int indexThreadCount = args.getInt("-indexThreadCount"); final String lineDocsFile = args.getString("-lineDocsFile"); final float docsPerSecPerThread = args.getFloat("-docsPerSecPerThread"); final float reopenEverySec = args.getFloat("-reopenEverySec"); final boolean storeBody = args.getFlag("-store"); final boolean tvsBody = args.getFlag("-tvs"); final boolean useCFS = args.getFlag("-cfs"); final String defaultPostingsFormat = args.getString("-postingsFormat"); final String idFieldPostingsFormat = args.getString("-idFieldPostingsFormat"); final boolean verbose = args.getFlag("-verbose"); final boolean cloneDocs = args.getFlag("-cloneDocs"); final Mode mode = Mode.valueOf(args.getString("-mode", "update").toUpperCase(Locale.ROOT)); final long reopenEveryMS = (long) (1000 * reopenEverySec); if (verbose) { InfoStream.setDefault(new PrintStreamInfoStream(System.out)); } if (!dirImpl.equals("RAMDirectory") && !dirImpl.equals("RAMExceptDirectPostingsDirectory")) { System.out.println("Wrap NRTCachingDirectory"); dir0 = new NRTCachingDirectory(dir0, 20, 400.0); } dir = dir0; final IndexWriterConfig iwc = new IndexWriterConfig(a); iwc.setOpenMode(IndexWriterConfig.OpenMode.APPEND); iwc.setRAMBufferSizeMB(256.0); iwc.setIndexDeletionPolicy(NoDeletionPolicy.INSTANCE); // TODO: also RAMDirExceptDirect...? need to // ... block deletes against wrapped FSDir? if (dirImpl.equals("RAMDirectory")) { // Let IW remove files only referenced by starting commit: iwc.setIndexDeletionPolicy(new KeepNoCommitsDeletionPolicy()); } if (commit != null && commit.length() > 0) { System.out.println("Opening writer on commit=" + commit); iwc.setIndexCommit(PerfUtils.findCommitPoint(commit, dir)); } ((TieredMergePolicy) iwc.getMergePolicy()).setNoCFSRatio(useCFS ? 1.0 : 0.0); //((TieredMergePolicy) iwc.getMergePolicy()).setMaxMergedSegmentMB(1024); //((TieredMergePolicy) iwc.getMergePolicy()).setReclaimDeletesWeight(3.0); //((TieredMergePolicy) iwc.getMergePolicy()).setMaxMergeAtOnce(4); final Codec codec = new Lucene62Codec() { @Override public PostingsFormat getPostingsFormatForField(String field) { return PostingsFormat .forName(field.equals("id") ? idFieldPostingsFormat : defaultPostingsFormat); } }; iwc.setCodec(codec); final ConcurrentMergeScheduler cms = (ConcurrentMergeScheduler) iwc.getMergeScheduler(); // Only let one merge run at a time... // ... but queue up up to 4, before index thread is stalled: cms.setMaxMergesAndThreads(4, 1); iwc.setMergedSegmentWarmer(new IndexWriter.IndexReaderWarmer() { @Override public void warm(LeafReader reader) throws IOException { final long t0 = System.currentTimeMillis(); //System.out.println("DO WARM: " + reader); IndexSearcher s = new IndexSearcher(reader); s.setQueryCache(null); // don't bench the cache s.search(new TermQuery(new Term(fieldName, "united")), 10); final long t1 = System.currentTimeMillis(); System.out.println("warm segment=" + reader + " numDocs=" + reader.numDocs() + ": took " + (t1 - t0) + " msec"); } }); writer = new IndexWriter(dir, iwc); System.out.println("Initial writer.maxDoc()=" + writer.maxDoc()); // TODO: add -nrtBodyPostingsOffsets instead of // hardwired false: boolean addDVFields = mode == Mode.BDV_UPDATE || mode == Mode.NDV_UPDATE; LineFileDocs lineFileDocs = new LineFileDocs(lineDocsFile, false, storeBody, tvsBody, false, cloneDocs, null, null, null, addDVFields); IndexThreads threads = new IndexThreads(new Random(17), writer, new AtomicBoolean(false), lineFileDocs, indexThreadCount, -1, false, false, mode, docsPerSecPerThread, null, -1.0, -1); threads.start(); mgr = new SearcherManager(writer, new SearcherFactory() { @Override public IndexSearcher newSearcher(IndexReader reader, IndexReader previous) { IndexSearcher s = new IndexSearcher(reader); s.setQueryCache(null); // don't bench the cache s.setSimilarity(sim); return s; } }); System.out.println("reopen every " + reopenEverySec); Thread reopenThread = new Thread() { @Override public void run() { try { final long startMS = System.currentTimeMillis(); int reopenCount = 1; while (true) { final long sleepMS = startMS + (reopenCount * reopenEveryMS) - System.currentTimeMillis(); if (sleepMS < 0) { System.out.println("WARNING: reopen fell behind by " + Math.abs(sleepMS) + " ms"); } else { Thread.sleep(sleepMS); } Thread.sleep(sleepMS); mgr.maybeRefresh(); reopenCount++; IndexSearcher s = mgr.acquire(); try { if (ramDir != null) { System.out.println(String.format(Locale.ENGLISH, "%.1fs: index: %d bytes in RAMDir; writer.maxDoc()=%d; searcher.maxDoc()=%d; searcher.numDocs()=%d", (System.currentTimeMillis() - startMS) / 1000.0, ramDir.ramBytesUsed(), writer.maxDoc(), s.getIndexReader().maxDoc(), s.getIndexReader().numDocs())); //String[] l = ramDir.listAll(); //Arrays.sort(l); //for(String f : l) { //System.out.println(" " + f + ": " + ramDir.fileLength(f)); //} } else { System.out.println(String.format(Locale.ENGLISH, "%.1fs: done reopen; writer.maxDoc()=%d; searcher.maxDoc()=%d; searcher.numDocs()=%d", (System.currentTimeMillis() - startMS) / 1000.0, writer.maxDoc(), s.getIndexReader().maxDoc(), s.getIndexReader().numDocs())); } } finally { mgr.release(s); } } } catch (Exception e) { throw new RuntimeException(e); } } }; reopenThread.setName("ReopenThread"); reopenThread.setPriority(4 + Thread.currentThread().getPriority()); reopenThread.start(); } else { dir = dir0; writer = null; final DirectoryReader reader; if (commit != null && commit.length() > 0) { System.out.println("Opening searcher on commit=" + commit); reader = DirectoryReader.open(PerfUtils.findCommitPoint(commit, dir)); } else { // open last commit reader = DirectoryReader.open(dir); } IndexSearcher s = new IndexSearcher(reader); s.setQueryCache(null); // don't bench the cache s.setSimilarity(sim); System.out.println("maxDoc=" + reader.maxDoc() + " numDocs=" + reader.numDocs() + " %tg deletes=" + (100. * reader.maxDoc() / reader.numDocs())); mgr = new SingleIndexSearcher(s); } System.out.println((System.currentTimeMillis() - tSearcherStart) + " msec to init searcher/NRT"); { IndexSearcher s = mgr.acquire(); try { System.out.println("Searcher: numDocs=" + s.getIndexReader().numDocs() + " maxDoc=" + s.getIndexReader().maxDoc() + ": " + s); } finally { mgr.release(s); } } //System.out.println("searcher=" + searcher); FacetsConfig facetsConfig = new FacetsConfig(); facetsConfig.setHierarchical("Date", true); TaxonomyReader taxoReader; Path taxoPath = Paths.get(args.getString("-indexPath"), "facets"); Directory taxoDir = od.open(taxoPath); if (DirectoryReader.indexExists(taxoDir)) { taxoReader = new DirectoryTaxonomyReader(taxoDir); System.out.println("Taxonomy has " + taxoReader.getSize() + " ords"); } else { taxoReader = null; } final Random staticRandom = new Random(staticRandomSeed); final Random random = new Random(randomSeed); final DirectSpellChecker spellChecker = new DirectSpellChecker(); final IndexState indexState = new IndexState(mgr, taxoReader, fieldName, spellChecker, hiliteImpl, facetsConfig); final QueryParser queryParser = new QueryParser("body", a); TaskParser taskParser = new TaskParser(indexState, queryParser, fieldName, topN, staticRandom, doStoredLoads); final TaskSource tasks; if (tasksFile.startsWith("server:")) { int idx = tasksFile.indexOf(':', 8); if (idx == -1) { throw new RuntimeException( "server is missing the port; should be server:interface:port (got: " + tasksFile + ")"); } String iface = tasksFile.substring(7, idx); int port = Integer.valueOf(tasksFile.substring(1 + idx)); RemoteTaskSource remoteTasks = new RemoteTaskSource(iface, port, searchThreadCount, taskParser); // nocommit must stop thread? tasks = remoteTasks; } else { // Load the tasks from a file: final int taskRepeatCount = args.getInt("-taskRepeatCount"); final int numTaskPerCat = args.getInt("-tasksPerCat"); tasks = new LocalTaskSource(indexState, taskParser, tasksFile, staticRandom, random, numTaskPerCat, taskRepeatCount, doPKLookup); System.out.println("Task repeat count " + taskRepeatCount); System.out.println("Tasks file " + tasksFile); System.out.println("Num task per cat " + numTaskPerCat); } args.check(); // Evil respeller: //spellChecker.setMinPrefix(0); //spellChecker.setMaxInspections(1024); final TaskThreads taskThreads = new TaskThreads(tasks, indexState, searchThreadCount); Thread.sleep(10); final long startNanos = System.nanoTime(); taskThreads.start(); taskThreads.finish(); final long endNanos = System.nanoTime(); System.out.println("\n" + ((endNanos - startNanos) / 1000000.0) + " msec total"); final List<Task> allTasks = tasks.getAllTasks(); PrintStream out = new PrintStream(logFile); if (allTasks != null) { // Tasks were local: verify checksums: // indexState.setDocIDToID(); final Map<Task, Task> tasksSeen = new HashMap<Task, Task>(); out.println("\nResults for " + allTasks.size() + " tasks:"); boolean fail = false; for (final Task task : allTasks) { if (verifyCheckSum) { final Task other = tasksSeen.get(task); if (other != null) { if (task.checksum() != other.checksum()) { System.out.println("\nTASK:"); task.printResults(System.out, indexState); System.out.println("\nOTHER TASK:"); other.printResults(System.out, indexState); fail = true; //throw new RuntimeException("task " + task + " hit different checksums: " + task.checksum() + " vs " + other.checksum() + " other=" + other); } } else { tasksSeen.put(task, task); } } out.println("\nTASK: " + task); out.println(" " + (task.runTimeNanos / 1000000.0) + " msec"); out.println(" thread " + task.threadID); task.printResults(out, indexState); } if (fail) { throw new RuntimeException("some tasks got different results across different threads"); } allTasks.clear(); } mgr.close(); if (taxoReader != null) { taxoReader.close(); } if (writer != null) { // Don't actually commit any index changes: writer.rollback(); } dir.close(); if (printHeap) { // Try to get RAM usage -- some ideas poached from http://www.javaworld.com/javaworld/javatips/jw-javatip130.html final Runtime runtime = Runtime.getRuntime(); long usedMem1 = PerfUtils.usedMemory(runtime); long usedMem2 = Long.MAX_VALUE; for (int iter = 0; iter < 10; iter++) { runtime.runFinalization(); runtime.gc(); Thread.yield(); Thread.sleep(100); usedMem2 = usedMem1; usedMem1 = PerfUtils.usedMemory(runtime); } out.println("\nHEAP: " + PerfUtils.usedMemory(runtime)); } out.close(); }