List of usage examples for org.apache.lucene.index IndexWriterConfig setMergePolicy
@Override
public IndexWriterConfig setMergePolicy(MergePolicy mergePolicy)
From source file:org.sonatype.nexus.index.NexusIndexingContext.java
License:Open Source License
@Override protected IndexWriterConfig getWriterConfig() { final IndexWriterConfig writerConfig = super.getWriterConfig(); // NEXUS-5380 force use of compound lucene index file to postpone "Too many open files" final TieredMergePolicy mergePolicy = new TieredMergePolicy(); mergePolicy.setUseCompoundFile(true); mergePolicy.setNoCFSRatio(1.0);/* ww w . ja v a 2 s.c o m*/ writerConfig.setMergePolicy(mergePolicy); return writerConfig; }
From source file:org.wikimedia.revdiffsearch.IndexMerger.java
License:Open Source License
/** Index all text files under a directory. */ public static void main(String[] args) { if (args.length != 2) { System.out.println("Usage: java -jar IndexMerger.jar " + "merged_index_dir existing_index_dir1 existing_index_dir2 ..."); System.out.println(" merged_index_dir: A directory where the merged " + "index will be stored"); System.out.println(" e.g. merged_indexes"); System.out//from w w w . j a v a 2s.co m .println(" existing_indexes_dir: A directory where the " + "indexes that have to merged exist"); System.out.println(" e.g. indexes/"); System.out.println(" e.g. index1"); System.out.println(" e.g. index2"); System.out.println(" e.g. index3"); System.exit(1); } int ramBufferSizeMB = 1024; int ngram = 3; File INDEX_DIR = new File(args[0]); { String s; if ((s = System.getProperty("ngram")) != null) { ngram = Integer.parseInt(s); } if ((s = System.getProperty("ramBufferSize")) != null) { ramBufferSizeMB = Integer.parseInt(s); } } INDEX_DIR.mkdir(); Date start = new Date(); try { IndexWriterConfig cfg = new IndexWriterConfig(Version.LUCENE_44, new SimpleNGramAnalyzer(ngram)); LogDocMergePolicy lmp = new LogDocMergePolicy(); lmp.setMergeFactor(1000); cfg.setRAMBufferSizeMB(ramBufferSizeMB); cfg.setMergePolicy(lmp); IndexWriter writer = new IndexWriter(FSDirectory.open(INDEX_DIR), cfg); // IndexWriter writer = new IndexWriter(INDEX_DIR, // new StandardAnalyzer(Version.LUCENE_44), // true); // writer.setMergeFactor(1000); // writer.setRAMBufferSizeMB(50); List<Directory> indexes = new ArrayList<Directory>(); for (String indexdir : Arrays.asList(args).subList(1, args.length)) { System.out.println("Adding: " + indexdir); indexes.add(FSDirectory.open(new File(indexdir))); } System.out.print("Merging added indexes..."); writer.addIndexes(indexes.toArray(new Directory[indexes.size()])); System.out.println("done"); writer.close(); System.out.println("done"); Date end = new Date(); System.out.println("It took: " + ((end.getTime() - start.getTime()) / 1000) + "\""); } catch (IOException e) { e.printStackTrace(); } }
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 {/*w w w . j ava 2s .c o m*/ 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.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;/*from www.j ava 2 s.c o 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.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 .j a va2 s.co m*/ } } 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.IndexGeoNames.java
License:Apache License
public static void main(String[] args) throws Exception { String geoNamesFile = args[0]; File indexPath = new File(args[1]); int numThreads = Integer.parseInt(args[2]); int precStep = Integer.parseInt(args[3]); if (indexPath.exists()) { throw new IllegalArgumentException("please remove indexPath \"" + indexPath + "\" before running"); }/* w w w .j ava 2 s . com*/ Directory dir = FSDirectory.open(indexPath); //IndexWriterConfig iwc = new IndexWriterConfig(Version.LUCENE_48, new StandardAnalyzer(Version.LUCENE_48)); IndexWriterConfig iwc = new IndexWriterConfig(new StandardAnalyzer()); //iwc.setRAMBufferSizeMB(350); iwc.setInfoStream(new PrintStreamInfoStream(System.out)); if (normal == false) { iwc.setRAMBufferSizeMB(1024); iwc.setMergePolicy(NoMergePolicy.INSTANCE); //iwc.setMergePolicy(NoMergePolicy.NO_COMPOUND_FILES); } else { // 5/5 segments: iwc.setMaxBufferedDocs(157234); iwc.setRAMBufferSizeMB(-1); } //((ConcurrentMergeScheduler) iwc.getMergeScheduler()).setMaxMergesAndThreads(3, 1); final IndexWriter w = new IndexWriter(dir, iwc); final Field.Store store = Field.Store.NO; final FieldType doubleFieldType = new FieldType( store == Field.Store.NO ? DoubleField.TYPE_NOT_STORED : DoubleField.TYPE_STORED); doubleFieldType.setNumericPrecisionStep(precStep); doubleFieldType.freeze(); final FieldType longFieldType = new FieldType( store == Field.Store.NO ? LongField.TYPE_NOT_STORED : LongField.TYPE_STORED); longFieldType.setNumericPrecisionStep(precStep); longFieldType.freeze(); final FieldType intFieldType = new FieldType( store == Field.Store.NO ? IntField.TYPE_NOT_STORED : IntField.TYPE_STORED); intFieldType.setNumericPrecisionStep(precStep); intFieldType.freeze(); // 64K buffer: InputStream is = new FileInputStream(geoNamesFile); final BufferedReader reader = new BufferedReader(new InputStreamReader(is, "UTF-8"), 1 << 16); final AtomicInteger docsIndexed = new AtomicInteger(); final long startMS = System.currentTimeMillis(); Thread[] threads = new Thread[numThreads]; // With reuse it's ~ 38% faster (41.8 sec vs 67.0 sec): final boolean reuseDocAndFields = false; for (int i = 0; i < numThreads; i++) { threads[i] = new Thread() { @Override public void run() { ParsePosition datePos = new ParsePosition(0); SimpleDateFormat dateParser = new SimpleDateFormat("yyyy-MM-dd", Locale.US); if (reuseDocAndFields) { Document doc = new Document(); IntField geoNameID = new IntField("geoNameID", 0, intFieldType); doc.add(geoNameID); TextField nameField = new TextField("name", "", store); doc.add(nameField); TextField asciiNameField = new TextField("asciiName", "", store); doc.add(asciiNameField); TextField alternateNameField = new TextField("alternateNames", "", store); doc.add(alternateNameField); StringField featureClassField = new StringField("featureClass", "", store); doc.add(featureClassField); StringField featureCodeField = new StringField("featureCode", "", store); doc.add(featureCodeField); StringField countryCodeField = new StringField("countryCode", "", store); doc.add(countryCodeField); StringField cc2Field = new StringField("cc2", "", store); doc.add(cc2Field); StringField admin1Field = new StringField("admin1", "", store); doc.add(admin1Field); StringField admin2Field = new StringField("admin2", "", store); doc.add(admin2Field); StringField admin3Field = new StringField("admin3", "", store); doc.add(admin3Field); StringField admin4Field = new StringField("admin4", "", store); doc.add(admin4Field); StringField tzField = new StringField("timezone", "", store); doc.add(tzField); while (true) { try { // Curiously BufferedReader.readLine seems to be thread-safe... String line = reader.readLine(); if (line == null) { break; } String[] values = line.split("\t"); geoNameID.setIntValue(Integer.parseInt(values[0])); nameField.setStringValue(values[1]); asciiNameField.setStringValue(values[2]); alternateNameField.setStringValue(values[3]); /* if (values[4].isEmpty() == false) { double v = Double.parseDouble(values[4]); doc.add(new DoubleField("latitude", v, doubleFieldType)); doc.add(new DoubleDocValuesField("latitude", v)); } if (values[5].isEmpty() == false) { double v = Double.parseDouble(values[5]); doc.add(new DoubleField("longitude", v, doubleFieldType)); doc.add(new DoubleDocValuesField("longitude", v)); } */ featureClassField.setStringValue(values[6]); featureCodeField.setStringValue(values[7]); countryCodeField.setStringValue(values[8]); cc2Field.setStringValue(values[9]); admin1Field.setStringValue(values[10]); admin2Field.setStringValue(values[11]); admin3Field.setStringValue(values[12]); admin4Field.setStringValue(values[13]); /* if (values[14].isEmpty() == false) { long v = Long.parseLong(values[14]); doc.add(new LongField("population", v, longFieldType)); doc.add(new NumericDocValuesField("population", v)); } if (values[15].isEmpty() == false) { long v = Long.parseLong(values[15]); doc.add(new LongField("elevation", v, longFieldType)); doc.add(new NumericDocValuesField("elevation", v)); } if (values[16].isEmpty() == false) { doc.add(new IntField("dem", Integer.parseInt(values[16]), intFieldType)); } */ tzField.setStringValue(values[17]); /* if (values[18].isEmpty() == false) { datePos.setIndex(0); Date date = dateParser.parse(values[18], datePos); doc.add(new LongField("modified", date.getTime(), longFieldType)); } */ w.addDocument(doc); int count = docsIndexed.incrementAndGet(); if (count % 200000 == 0) { long ms = System.currentTimeMillis(); System.out.println(count + ": " + ((ms - startMS) / 1000.0) + " sec"); } } catch (Exception e) { throw new RuntimeException(e); } } } else { while (true) { try { // Curiously BufferedReader.readLine seems to be thread-safe... String line = reader.readLine(); if (line == null) { break; } String[] values = line.split("\t"); Document doc = new Document(); doc.add(new IntField("geoNameID", Integer.parseInt(values[0]), intFieldType)); doc.add(new TextField("name", values[1], store)); doc.add(new TextField("asciiName", values[2], store)); doc.add(new TextField("alternateNames", values[3], store)); if (values[4].isEmpty() == false) { double v = Double.parseDouble(values[4]); doc.add(new DoubleField("latitude", v, doubleFieldType)); doc.add(new DoubleDocValuesField("latitude", v)); } if (values[5].isEmpty() == false) { double v = Double.parseDouble(values[5]); doc.add(new DoubleField("longitude", v, doubleFieldType)); doc.add(new DoubleDocValuesField("longitude", v)); } doc.add(new StringField("featureClass", values[6], store)); doc.add(new StringField("featureCode", values[7], store)); doc.add(new StringField("countryCode", values[8], store)); doc.add(new StringField("cc2", values[9], store)); doc.add(new StringField("admin1Code", values[10], store)); doc.add(new StringField("admin2Code", values[11], store)); doc.add(new StringField("admin3Code", values[12], store)); doc.add(new StringField("admin4Code", values[13], store)); if (values[14].isEmpty() == false) { long v = Long.parseLong(values[14]); doc.add(new LongField("population", v, longFieldType)); doc.add(new NumericDocValuesField("population", v)); } if (values[15].isEmpty() == false) { long v = Long.parseLong(values[15]); doc.add(new LongField("elevation", v, longFieldType)); doc.add(new NumericDocValuesField("elevation", v)); } if (values[16].isEmpty() == false) { doc.add(new IntField("dem", Integer.parseInt(values[16]), intFieldType)); } doc.add(new StringField("timezone", values[17], store)); if (values[18].isEmpty() == false) { datePos.setIndex(0); Date date = dateParser.parse(values[18], datePos); doc.add(new LongField("modified", date.getTime(), longFieldType)); } w.addDocument(doc); int count = docsIndexed.incrementAndGet(); if (count % 200000 == 0) { long ms = System.currentTimeMillis(); System.out.println(count + ": " + ((ms - startMS) / 1000.0) + " sec"); } } catch (Exception e) { throw new RuntimeException(e); } } } } }; threads[i].start(); } DirectoryReader r = DirectoryReader.open(w, true); for (int i = 0; i < 100; i++) { DirectoryReader r2 = DirectoryReader.openIfChanged(r); if (r2 != null) { r.close(); r = r2; } Thread.sleep(500); } if (r != null) { r.close(); r = null; } for (int i = 0; i < numThreads; i++) { threads[i].join(); } long ms = System.currentTimeMillis(); System.out.println(docsIndexed + ": " + ((ms - startMS) / 1000.0) + " sec"); //System.out.println("tot conflicts: " + BytesRefHash.totConflict); //w.shutdown(normal); w.close(); dir.close(); }
From source file:perf.IndexNumbers.java
License:Apache License
public static void main(String[] args) throws Exception { String numbersFile = args[0]; File indexPath = new File(args[1]); int numThreads = Integer.parseInt(args[2]); int precStep = Integer.parseInt(args[3]); if (indexPath.exists()) { throw new IllegalArgumentException("please remove indexPath \"" + indexPath + "\" before running"); }/*from ww w . ja va 2s .co m*/ Directory dir = FSDirectory.open(indexPath); IndexWriterConfig iwc = new IndexWriterConfig(new StandardAnalyzer()); //iwc.setRAMBufferSizeMB(350); //iwc.setInfoStream(new PrintStreamInfoStream(System.out)); if (normal == false) { iwc.setRAMBufferSizeMB(64); iwc.setMergePolicy(NoMergePolicy.INSTANCE); } else { // 5/5 segments: // 3273681 docs in twitter timestamps iwc.setMaxBufferedDocs(59522); iwc.setRAMBufferSizeMB(-1); } //((ConcurrentMergeScheduler) iwc.getMergeScheduler()).setMaxMergesAndThreads(3, 1); final IndexWriter w = new IndexWriter(dir, iwc); final Field.Store store = Field.Store.NO; final FieldType doubleFieldType = new FieldType( store == Field.Store.NO ? DoubleField.TYPE_NOT_STORED : DoubleField.TYPE_STORED); doubleFieldType.setNumericPrecisionStep(precStep); doubleFieldType.freeze(); final FieldType longFieldType = new FieldType( store == Field.Store.NO ? LongField.TYPE_NOT_STORED : LongField.TYPE_STORED); longFieldType.setNumericPrecisionStep(precStep); longFieldType.freeze(); final FieldType intFieldType = new FieldType( store == Field.Store.NO ? IntField.TYPE_NOT_STORED : IntField.TYPE_STORED); intFieldType.setNumericPrecisionStep(precStep); intFieldType.freeze(); // 64K buffer: InputStream is = new FileInputStream(numbersFile); final BufferedReader reader = new BufferedReader(new InputStreamReader(is, "UTF-8"), 1 << 16); final AtomicInteger docsIndexed = new AtomicInteger(); final long startMS = System.currentTimeMillis(); Thread[] threads = new Thread[numThreads]; for (int i = 0; i < numThreads; i++) { threads[i] = new Thread() { @Override public void run() { ParsePosition datePos = new ParsePosition(0); SimpleDateFormat dateParser = new SimpleDateFormat("yyyy-MM-dd", Locale.US); Document doc = new Document(); Field field = new LongField("number", 0L, longFieldType); doc.add(field); while (true) { try { // Curiously BufferedReader.readLine seems to be thread-safe... String line = reader.readLine(); if (line == null) { break; } field.setLongValue(Long.parseLong(line.trim())); w.addDocument(doc); int count = docsIndexed.incrementAndGet(); if (count % 200000 == 0) { long ms = System.currentTimeMillis(); System.out.println(count + ": " + ((ms - startMS) / 1000.0) + " sec"); } } catch (Exception e) { throw new RuntimeException(e); } } } }; threads[i].start(); } for (int i = 0; i < numThreads; i++) { threads[i].join(); } long ms = System.currentTimeMillis(); System.out.println(docsIndexed + ": " + ((ms - startMS) / 1000.0) + " sec"); //System.out.println("tot conflicts: " + BytesRefHash.totConflict); if (normal == false) { w.abortMerges(); } w.close(); dir.close(); }
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 w w . j a v a 2 s. c o m 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:perLucene.Server.java
License:Open Source License
private static void initIndexWriter() { try {/*from w w w . j av a 2s. c o m*/ Directory dir = new MMapDirectory(new File("/mnt/perLucene")); // FSDirectory.setMaxMergeWriteMBPerSec. // use this if merges interfere with the search // use a unix native directory to tell the os not to cache merges NRTCachingDirectory index = new NRTCachingDirectory(dir, 500, 1000); IndexWriterConfig config = new IndexWriterConfig(Version.LUCENE_41, null); // used to reduce latency of obtaining a reader after a big merge // config.setMergedSegmentWarmer(new // IndexWriter.IndexReaderWarmer()); config.setMergePolicy(new LogByteSizeMergePolicy()); w = new IndexWriter(index, config); } catch (IOException e) { System.out.println("Stacktrace " + e.toString()); } }
From source file:proj.zoie.impl.indexing.internal.DiskSearchIndex.java
License:Apache License
/** * Opens an index modifier./*from w ww . ja v a 2 s. c o m*/ * @param analyzer Analyzer * @return IndexModifer instance */ public IndexWriter openIndexWriter(Analyzer analyzer, Similarity similarity) throws IOException { if (_indexWriter != null) return _indexWriter; Directory directory = _dirMgr.getDirectory(true); log.info("opening index writer at: " + _dirMgr.getPath()); ZoieMergePolicy mergePolicy = new ZoieMergePolicy(); mergePolicy.setMergePolicyParams(_mergePolicyParams); // hao: autocommit is set to false with this constructor IndexWriterConfig config = new IndexWriterConfig(Version.LUCENE_34, analyzer); config.setOpenMode(OpenMode.CREATE_OR_APPEND); config.setIndexDeletionPolicy(_deletionPolicy); config.setMergeScheduler(_mergeScheduler); config.setMergePolicy(mergePolicy); config.setReaderPooling(false); if (similarity != null) { config.setSimilarity(similarity); } config.setRAMBufferSizeMB(5); IndexWriter idxWriter = new IndexWriter(directory, config); _indexWriter = idxWriter; return idxWriter; }