internal virtual void SealFlushedSegment(FlushedSegment flushedSegment)
        {
            if (Debugging.AssertsEnabled)
            {
                Debugging.Assert(flushedSegment != null);
            }

            SegmentCommitInfo newSegment = flushedSegment.segmentInfo;

            IndexWriter.SetDiagnostics(newSegment.Info, IndexWriter.SOURCE_FLUSH);

            IOContext context = new IOContext(new FlushInfo(newSegment.Info.DocCount, newSegment.GetSizeInBytes()));

            bool success = false;

            try
            {
                if (indexWriterConfig.UseCompoundFile)
                {
                    filesToDelete.UnionWith(IndexWriter.CreateCompoundFile(infoStream, directory, CheckAbort.NONE, newSegment.Info, context));
                    newSegment.Info.UseCompoundFile = true;
                }

                // 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:
                codec.SegmentInfoFormat.SegmentInfoWriter.Write(directory, newSegment.Info, flushedSegment.fieldInfos, context);

                // TODO: ideally we would freeze newSegment here!!
                // because any changes after writing the .si will be
                // lost...

                // Must write deleted docs after the CFS so we don't
                // slurp the del file into CFS:
                if (flushedSegment.liveDocs != null)
                {
                    int delCount = flushedSegment.delCount;
                    if (Debugging.AssertsEnabled)
                    {
                        Debugging.Assert(delCount > 0);
                    }
                    if (infoStream.IsEnabled("DWPT"))
                    {
                        infoStream.Message("DWPT", "flush: write " + delCount + " deletes gen=" + flushedSegment.segmentInfo.DelGen);
                    }

                    // TODO: we should prune the segment if it's 100%
                    // deleted... but merge will also catch it.

                    // TODO: in the NRT case it'd be better to hand
                    // this del vector over to the
                    // shortly-to-be-opened SegmentReader and let it
                    // carry the changes; there's no reason to use
                    // filesystem as intermediary here.

                    SegmentCommitInfo info  = flushedSegment.segmentInfo;
                    Codec             codec = info.Info.Codec;
                    codec.LiveDocsFormat.WriteLiveDocs(flushedSegment.liveDocs, directory, info, delCount, context);
                    newSegment.DelCount = delCount;
                    newSegment.AdvanceDelGen();
                }

                success = true;
            }
            finally
            {
                if (!success)
                {
                    if (infoStream.IsEnabled("DWPT"))
                    {
                        infoStream.Message("DWPT", "hit exception creating compound file for newly flushed segment " + newSegment.Info.Name);
                    }
                }
            }
        }
Exemple #2
0
 public virtual void ListSegments()
 {
     for (int x = 0; x < Infos.Count; x++)
     {
         SegmentCommitInfo info    = Infos.Info(x);
         string            sizeStr = string.Format(CultureInfo.InvariantCulture, "{0:###,###.###}", info.GetSizeInBytes());
         Console.WriteLine(info.Info.Name + " " + sizeStr);
     }
 }
        internal virtual FlushedSegment Flush()
        {
            if (Debugging.AssertsEnabled)
            {
                Debugging.Assert(numDocsInRAM > 0);
                Debugging.Assert(deleteSlice.IsEmpty, "all deletes must be applied in prepareFlush");
            }
            segmentInfo.DocCount = numDocsInRAM;
            SegmentWriteState flushState  = new SegmentWriteState(infoStream, directory, segmentInfo, fieldInfos.Finish(), indexWriterConfig.TermIndexInterval, pendingUpdates, new IOContext(new FlushInfo(numDocsInRAM, BytesUsed)));
            double            startMBUsed = BytesUsed / 1024.0 / 1024.0;

            // Apply delete-by-docID now (delete-byDocID only
            // happens when an exception is hit processing that
            // doc, eg if analyzer has some problem w/ the text):
            if (pendingUpdates.docIDs.Count > 0)
            {
                flushState.LiveDocs = codec.LiveDocsFormat.NewLiveDocs(numDocsInRAM);
                foreach (int delDocID in pendingUpdates.docIDs)
                {
                    flushState.LiveDocs.Clear(delDocID);
                }
                flushState.DelCountOnFlush = pendingUpdates.docIDs.Count;
                pendingUpdates.bytesUsed.AddAndGet(-pendingUpdates.docIDs.Count * BufferedUpdates.BYTES_PER_DEL_DOCID);
                pendingUpdates.docIDs.Clear();
            }

            if (aborting)
            {
                if (infoStream.IsEnabled("DWPT"))
                {
                    infoStream.Message("DWPT", "flush: skip because aborting is set");
                }
                return(null);
            }

            if (infoStream.IsEnabled("DWPT"))
            {
                infoStream.Message("DWPT", "flush postings as segment " + flushState.SegmentInfo.Name + " numDocs=" + numDocsInRAM);
            }

            bool success = false;

            try
            {
                consumer.Flush(flushState);
                pendingUpdates.terms.Clear();
                segmentInfo.SetFiles(new JCG.HashSet <string>(directory.CreatedFiles));

                SegmentCommitInfo segmentInfoPerCommit = new SegmentCommitInfo(segmentInfo, 0, -1L, -1L);
                if (infoStream.IsEnabled("DWPT"))
                {
                    infoStream.Message("DWPT", "new segment has " + (flushState.LiveDocs == null ? 0 : (flushState.SegmentInfo.DocCount - flushState.DelCountOnFlush)) + " deleted docs");
                    infoStream.Message("DWPT", "new segment has " + (flushState.FieldInfos.HasVectors ? "vectors" : "no vectors") + "; " + (flushState.FieldInfos.HasNorms ? "norms" : "no norms") + "; " + (flushState.FieldInfos.HasDocValues ? "docValues" : "no docValues") + "; " + (flushState.FieldInfos.HasProx ? "prox" : "no prox") + "; " + (flushState.FieldInfos.HasFreq ? "freqs" : "no freqs"));
                    infoStream.Message("DWPT", "flushedFiles=" + string.Format(J2N.Text.StringFormatter.InvariantCulture, "{0}", segmentInfoPerCommit.GetFiles()));
                    infoStream.Message("DWPT", "flushed codec=" + codec);
                }

                BufferedUpdates segmentDeletes;
                if (pendingUpdates.queries.Count == 0 && pendingUpdates.numericUpdates.Count == 0 && pendingUpdates.binaryUpdates.Count == 0)
                {
                    pendingUpdates.Clear();
                    segmentDeletes = null;
                }
                else
                {
                    segmentDeletes = pendingUpdates;
                }

                if (infoStream.IsEnabled("DWPT"))
                {
                    double newSegmentSize = segmentInfoPerCommit.GetSizeInBytes() / 1024.0 / 1024.0;
                    infoStream.Message("DWPT", "flushed: segment=" + segmentInfo.Name + " ramUsed=" + startMBUsed.ToString(nf) + " MB" + " newFlushedSize(includes docstores)=" + newSegmentSize.ToString(nf) + " MB" + " docs/MB=" + (flushState.SegmentInfo.DocCount / newSegmentSize).ToString(nf));
                }

                if (Debugging.AssertsEnabled)
                {
                    Debugging.Assert(segmentInfo != null);
                }

                FlushedSegment fs = new FlushedSegment(segmentInfoPerCommit, flushState.FieldInfos, segmentDeletes, flushState.LiveDocs, flushState.DelCountOnFlush);
                SealFlushedSegment(fs);
                success = true;

                return(fs);
            }
            finally
            {
                if (!success)
                {
                    Abort(filesToDelete);
                }
            }
        }