Examples with InfoStream org.apache.lucene.util.InfoStream used on opensource projects

Example 1 with InfoStream

use of org.apache.lucene.util.InfoStream in project lucene-solr by apache.

the class CloseIndexTask method doLogic.

@Override
public int doLogic() throws IOException {
    IndexWriter iw = getRunData().getIndexWriter();
    if (iw != null) {
        // If infoStream was set to output to a file, close it.
        InfoStream infoStream = iw.getConfig().getInfoStream();
        if (infoStream != null) {
            infoStream.close();
        }
        if (doWait == false) {
            iw.commit();
            iw.rollback();
        } else {
            iw.close();
        }
        getRunData().setIndexWriter(null);
    }
    return 1;
}

Example 2 with InfoStream

use of org.apache.lucene.util.InfoStream in project lucene-solr by apache.

the class TestInfoStream method testTestPointsOn.

/** but they should work when we need */
public void testTestPointsOn() throws Exception {
    Directory dir = newDirectory();
    IndexWriterConfig iwc = new IndexWriterConfig(null);
    AtomicBoolean seenTestPoint = new AtomicBoolean();
    iwc.setInfoStream(new InfoStream() {

        @Override
        public void close() throws IOException {
        }

        @Override
        public void message(String component, String message) {
            if ("TP".equals(component)) {
                seenTestPoint.set(true);
            }
        }

        @Override
        public boolean isEnabled(String component) {
            return true;
        }
    });
    IndexWriter iw = new IndexWriter(dir, iwc);
    iw.enableTestPoints = true;
    iw.addDocument(new Document());
    iw.close();
    dir.close();
    assertTrue(seenTestPoint.get());
}

Example 3 with InfoStream

use of org.apache.lucene.util.InfoStream in project lucene-solr by apache.

the class TestIndexWriterReader method testSimpleMergedSegmentWarmer.

public void testSimpleMergedSegmentWarmer() throws Exception {
    Directory dir = newDirectory();
    final AtomicBoolean didWarm = new AtomicBoolean();
    InfoStream infoStream = new InfoStream() {

        @Override
        public void close() throws IOException {
        }

        @Override
        public void message(String component, String message) {
            if ("SMSW".equals(component)) {
                didWarm.set(true);
            }
        }

        @Override
        public boolean isEnabled(String component) {
            return true;
        }
    };
    IndexWriter w = new IndexWriter(dir, newIndexWriterConfig(new MockAnalyzer(random())).setMaxBufferedDocs(2).setReaderPooling(true).setInfoStream(infoStream).setMergedSegmentWarmer(new SimpleMergedSegmentWarmer(infoStream)).setMergePolicy(newLogMergePolicy(10)));
    Document doc = new Document();
    doc.add(newStringField("foo", "bar", Field.Store.NO));
    for (int i = 0; i < 20; i++) {
        w.addDocument(doc);
    }
    w.waitForMerges();
    w.close();
    dir.close();
    assertTrue(didWarm.get());
}

Example 4 with InfoStream

use of org.apache.lucene.util.InfoStream in project lucene-solr by apache.

the class IndexWriter method mergeMiddle.

/** Does the actual (time-consuming) work of the merge,
   *  but without holding synchronized lock on IndexWriter
   *  instance */
private int mergeMiddle(MergePolicy.OneMerge merge, MergePolicy mergePolicy) throws IOException {
    merge.checkAborted();
    Directory mergeDirectory = config.getMergeScheduler().wrapForMerge(merge, directory);
    List<SegmentCommitInfo> sourceSegments = merge.segments;
    IOContext context = new IOContext(merge.getStoreMergeInfo());
    final TrackingDirectoryWrapper dirWrapper = new TrackingDirectoryWrapper(mergeDirectory);
    if (infoStream.isEnabled("IW")) {
        infoStream.message("IW", "merging " + segString(merge.segments));
    }
    merge.readers = new ArrayList<>(sourceSegments.size());
    // This is try/finally to make sure merger's readers are
    // closed:
    boolean success = false;
    try {
        int segUpto = 0;
        while (segUpto < sourceSegments.size()) {
            final SegmentCommitInfo info = sourceSegments.get(segUpto);
            // Hold onto the "live" reader; we will use this to
            // commit merged deletes
            final ReadersAndUpdates rld = readerPool.get(info, true);
            // Carefully pull the most recent live docs and reader
            SegmentReader reader;
            final Bits liveDocs;
            final int delCount;
            synchronized (this) {
                // Must sync to ensure BufferedDeletesStream cannot change liveDocs,
                // pendingDeleteCount and field updates while we pull a copy:
                reader = rld.getReaderForMerge(context);
                liveDocs = rld.getReadOnlyLiveDocs();
                delCount = rld.getPendingDeleteCount() + info.getDelCount();
                assert reader != null;
                assert rld.verifyDocCounts();
                if (infoStream.isEnabled("IW")) {
                    if (rld.getPendingDeleteCount() != 0) {
                        infoStream.message("IW", "seg=" + segString(info) + " delCount=" + info.getDelCount() + " pendingDelCount=" + rld.getPendingDeleteCount());
                    } else if (info.getDelCount() != 0) {
                        infoStream.message("IW", "seg=" + segString(info) + " delCount=" + info.getDelCount());
                    } else {
                        infoStream.message("IW", "seg=" + segString(info) + " no deletes");
                    }
                }
            }
            // make a new reader with updated live docs and del count.
            if (reader.numDeletedDocs() != delCount) {
                // beware of zombies
                assert delCount > reader.numDeletedDocs();
                SegmentReader newReader;
                synchronized (this) {
                    // We must also sync on IW here, because another thread could be writing
                    // new DV updates / remove old gen field infos files causing FNFE:
                    newReader = new SegmentReader(info, reader, liveDocs, info.info.maxDoc() - delCount);
                }
                boolean released = false;
                try {
                    rld.release(reader);
                    released = true;
                } finally {
                    if (!released) {
                        newReader.decRef();
                    }
                }
                reader = newReader;
            }
            merge.readers.add(reader);
            assert delCount <= info.info.maxDoc() : "delCount=" + delCount + " info.maxDoc=" + info.info.maxDoc() + " rld.pendingDeleteCount=" + rld.getPendingDeleteCount() + " info.getDelCount()=" + info.getDelCount();
            segUpto++;
        }
        //      System.out.println("[" + Thread.currentThread().getName() + "] IW.mergeMiddle: merging " + merge.getMergeReaders());
        // Let the merge wrap readers
        List<CodecReader> mergeReaders = new ArrayList<>();
        for (SegmentReader reader : merge.readers) {
            CodecReader wrappedReader = merge.wrapForMerge(reader);
            validateMergeReader(wrappedReader);
            mergeReaders.add(wrappedReader);
        }
        final SegmentMerger merger = new SegmentMerger(mergeReaders, merge.info.info, infoStream, dirWrapper, globalFieldNumberMap, context);
        merge.checkAborted();
        merge.mergeStartNS = System.nanoTime();
        // This is where all the work happens:
        if (merger.shouldMerge()) {
            merger.merge();
        }
        MergeState mergeState = merger.mergeState;
        assert mergeState.segmentInfo == merge.info.info;
        merge.info.info.setFiles(new HashSet<>(dirWrapper.getCreatedFiles()));
        if (infoStream.isEnabled("IW")) {
            if (merger.shouldMerge()) {
                String pauseInfo = merge.getMergeProgress().getPauseTimes().entrySet().stream().filter((e) -> e.getValue() > 0).map((e) -> String.format(Locale.ROOT, "%.1f sec %s", e.getValue() / 1000000000., e.getKey().name().toLowerCase(Locale.ROOT))).collect(Collectors.joining(", "));
                if (!pauseInfo.isEmpty()) {
                    pauseInfo = " (" + pauseInfo + ")";
                }
                long t1 = System.nanoTime();
                double sec = (t1 - merge.mergeStartNS) / 1000000000.;
                double segmentMB = (merge.info.sizeInBytes() / 1024. / 1024.);
                infoStream.message("IW", "merge codec=" + codec + " maxDoc=" + merge.info.info.maxDoc() + "; merged segment has " + (mergeState.mergeFieldInfos.hasVectors() ? "vectors" : "no vectors") + "; " + (mergeState.mergeFieldInfos.hasNorms() ? "norms" : "no norms") + "; " + (mergeState.mergeFieldInfos.hasDocValues() ? "docValues" : "no docValues") + "; " + (mergeState.mergeFieldInfos.hasProx() ? "prox" : "no prox") + "; " + (mergeState.mergeFieldInfos.hasProx() ? "freqs" : "no freqs") + "; " + (mergeState.mergeFieldInfos.hasPointValues() ? "points" : "no points") + "; " + String.format(Locale.ROOT, "%.1f sec%s to merge segment [%.2f MB, %.2f MB/sec]", sec, pauseInfo, segmentMB, segmentMB / sec));
            } else {
                infoStream.message("IW", "skip merging fully deleted segments");
            }
        }
        if (merger.shouldMerge() == false) {
            // Merge would produce a 0-doc segment, so we do nothing except commit the merge to remove all the 0-doc segments that we "merged":
            assert merge.info.info.maxDoc() == 0;
            commitMerge(merge, mergeState);
            return 0;
        }
        assert merge.info.info.maxDoc() > 0;
        // Very important to do this before opening the reader
        // because codec must know if prox was written for
        // this segment:
        //System.out.println("merger set hasProx=" + merger.hasProx() + " seg=" + merge.info.name);
        boolean useCompoundFile;
        synchronized (this) {
            // Guard segmentInfos
            useCompoundFile = mergePolicy.useCompoundFile(segmentInfos, merge.info, this);
        }
        if (useCompoundFile) {
            success = false;
            Collection<String> filesToRemove = merge.info.files();
            TrackingDirectoryWrapper trackingCFSDir = new TrackingDirectoryWrapper(mergeDirectory);
            try {
                createCompoundFile(infoStream, trackingCFSDir, merge.info.info, context);
                success = true;
            } catch (Throwable t) {
                synchronized (this) {
                    if (merge.isAborted()) {
                        // merged files:
                        if (infoStream.isEnabled("IW")) {
                            infoStream.message("IW", "hit merge abort exception creating compound file during merge");
                        }
                        return 0;
                    } else {
                        handleMergeException(t, merge);
                    }
                }
            } finally {
                if (success == false) {
                    if (infoStream.isEnabled("IW")) {
                        infoStream.message("IW", "hit exception creating compound file during merge");
                    }
                    // Safe: these files must exist
                    deleteNewFiles(merge.info.files());
                }
            }
            // So that, if we hit exc in deleteNewFiles (next)
            // or in commitMerge (later), we close the
            // per-segment readers in the finally clause below:
            success = false;
            synchronized (this) {
                // delete new non cfs files directly: they were never
                // registered with IFD
                deleteNewFiles(filesToRemove);
                if (merge.isAborted()) {
                    if (infoStream.isEnabled("IW")) {
                        infoStream.message("IW", "abort merge after building CFS");
                    }
                    // Safe: these files must exist
                    deleteNewFiles(merge.info.files());
                    return 0;
                }
            }
            merge.info.info.setUseCompoundFile(true);
        } else {
            // So that, if we hit exc in commitMerge (later),
            // we close the per-segment readers in the finally
            // clause below:
            success = false;
        }
        // Have codec write SegmentInfo.  Must do this after
        // creating CFS so that 1) .si isn't slurped into CFS,
        // and 2) .si reflects useCompoundFile=true change
        // above:
        boolean success2 = false;
        try {
            codec.segmentInfoFormat().write(directory, merge.info.info, context);
            success2 = true;
        } finally {
            if (!success2) {
                // Safe: these files must exist
                deleteNewFiles(merge.info.files());
            }
        }
        if (infoStream.isEnabled("IW")) {
            infoStream.message("IW", String.format(Locale.ROOT, "merged segment size=%.3f MB vs estimate=%.3f MB", merge.info.sizeInBytes() / 1024. / 1024., merge.estimatedMergeBytes / 1024 / 1024.));
        }
        final IndexReaderWarmer mergedSegmentWarmer = config.getMergedSegmentWarmer();
        if (poolReaders && mergedSegmentWarmer != null) {
            final ReadersAndUpdates rld = readerPool.get(merge.info, true);
            final SegmentReader sr = rld.getReader(IOContext.READ);
            try {
                mergedSegmentWarmer.warm(sr);
            } finally {
                synchronized (this) {
                    rld.release(sr);
                    readerPool.release(rld);
                }
            }
        }
        if (!commitMerge(merge, mergeState)) {
            // aborted
            return 0;
        }
        success = true;
    } finally {
        // an exc:
        if (success == false) {
            closeMergeReaders(merge, true);
        }
    }
    return merge.info.info.maxDoc();
}

Example 5 with InfoStream

use of org.apache.lucene.util.InfoStream in project lucene-solr by apache.

the class IndexUpgrader method parseArgs.

@SuppressForbidden(reason = "System.out required: command line tool")
static IndexUpgrader parseArgs(String[] args) throws IOException {
    String path = null;
    boolean deletePriorCommits = false;
    InfoStream out = null;
    String dirImpl = null;
    int i = 0;
    while (i < args.length) {
        String arg = args[i];
        if ("-delete-prior-commits".equals(arg)) {
            deletePriorCommits = true;
        } else if ("-verbose".equals(arg)) {
            out = new PrintStreamInfoStream(System.out);
        } else if ("-dir-impl".equals(arg)) {
            if (i == args.length - 1) {
                System.out.println("ERROR: missing value for -dir-impl option");
                System.exit(1);
            }
            i++;
            dirImpl = args[i];
        } else if (path == null) {
            path = arg;
        } else {
            printUsage();
        }
        i++;
    }
    if (path == null) {
        printUsage();
    }
    Path p = Paths.get(path);
    Directory dir = null;
    if (dirImpl == null) {
        dir = FSDirectory.open(p);
    } else {
        dir = CommandLineUtil.newFSDirectory(dirImpl, p);
    }
    return new IndexUpgrader(dir, out, deletePriorCommits);
}