/// <summary>
/// Does the actual (time-consuming) work of the merge,
/// but without holding synchronized lock on IndexWriter
/// instance
/// </summary>
private int MergeMiddle(MergePolicy.OneMerge merge)
{
merge.CheckAborted(directory);
string mergedName = merge.Info_Renamed.Info.Name;
IList<SegmentCommitInfo> sourceSegments = merge.Segments;
IOContext context = new IOContext(merge.MergeInfo);
MergeState.CheckAbort checkAbort = new MergeState.CheckAbort(merge, directory);
TrackingDirectoryWrapper dirWrapper = new TrackingDirectoryWrapper(directory);
if (infoStream.IsEnabled("IW"))
{
infoStream.Message("IW", "merging " + SegString(merge.Segments));
}
merge.Readers = new List<SegmentReader>();
// this is try/finally to make sure merger's readers are
// closed:
bool success = false;
try
{
int segUpto = 0;
while (segUpto < sourceSegments.Count)
{
SegmentCommitInfo info = sourceSegments[segUpto];
// Hold onto the "live" reader; we will use this to
// commit merged deletes
ReadersAndUpdates rld = readerPool.Get(info, true);
// Carefully pull the most recent live docs and reader
SegmentReader reader;
Bits liveDocs;
int delCount;
lock (this)
{
// Must sync to ensure BufferedDeletesStream cannot change liveDocs,
// pendingDeleteCount and field updates while we pull a copy:
reader = rld.GetReaderForMerge(context);
liveDocs = rld.ReadOnlyLiveDocs;
delCount = rld.PendingDeleteCount + info.DelCount;
Debug.Assert(reader != null);
Debug.Assert(rld.VerifyDocCounts());
if (infoStream.IsEnabled("IW"))
{
if (rld.PendingDeleteCount != 0)
{
infoStream.Message("IW", "seg=" + SegString(info) + " delCount=" + info.DelCount + " pendingDelCount=" + rld.PendingDeleteCount);
}
else if (info.DelCount != 0)
{
infoStream.Message("IW", "seg=" + SegString(info) + " delCount=" + info.DelCount);
}
else
{
infoStream.Message("IW", "seg=" + SegString(info) + " no deletes");
}
}
}
// Deletes might have happened after we pulled the merge reader and
// before we got a read-only copy of the segment's actual live docs
// (taking pending deletes into account). In that case we need to
// make a new reader with updated live docs and del count.
if (reader.NumDeletedDocs() != delCount)
{
// fix the reader's live docs and del count
Debug.Assert(delCount > reader.NumDeletedDocs()); // beware of zombies
SegmentReader newReader = new SegmentReader(info, reader, liveDocs, info.Info.DocCount - delCount);
bool released = false;
try
{
rld.Release(reader);
released = true;
}
finally
{
if (!released)
{
newReader.DecRef();
}
}
reader = newReader;
}
merge.Readers.Add(reader);
Debug.Assert(delCount <= info.Info.DocCount, "delCount=" + delCount + " info.docCount=" + info.Info.DocCount + " rld.pendingDeleteCount=" + rld.PendingDeleteCount + " info.getDelCount()=" + info.DelCount);
segUpto++;
}
// System.out.println("[" + Thread.currentThread().getName() + "] IW.mergeMiddle: merging " + merge.getMergeReaders());
// we pass merge.getMergeReaders() instead of merge.readers to allow the
// OneMerge to return a view over the actual segments to merge
SegmentMerger merger = new SegmentMerger(merge.MergeReaders, merge.Info_Renamed.Info, infoStream, dirWrapper, Config_Renamed.TermIndexInterval, checkAbort, GlobalFieldNumberMap, context, Config_Renamed.CheckIntegrityAtMerge);
merge.CheckAborted(directory);
// this is where all the work happens:
MergeState mergeState;
bool success3 = false;
try
{
if (!merger.ShouldMerge())
{
// would result in a 0 document segment: nothing to merge!
mergeState = new MergeState(new List<AtomicReader>(), merge.Info_Renamed.Info, infoStream, checkAbort);
}
else
{
mergeState = merger.Merge();
}
success3 = true;
}
finally
{
if (!success3)
{
lock (this)
{
Deleter.Refresh(merge.Info_Renamed.Info.Name);
}
}
}
Debug.Assert(mergeState.SegmentInfo == merge.Info_Renamed.Info);
merge.Info_Renamed.Info.Files = new HashSet<string>(dirWrapper.CreatedFiles);
// Record which codec was used to write the segment
if (infoStream.IsEnabled("IW"))
{
if (merge.Info_Renamed.Info.DocCount == 0)
{
infoStream.Message("IW", "merge away fully deleted segments");
}
else
{
infoStream.Message("IW", "merge codec=" + Codec + " docCount=" + merge.Info_Renamed.Info.DocCount + "; merged segment has " + (mergeState.FieldInfos.HasVectors() ? "vectors" : "no vectors") + "; " + (mergeState.FieldInfos.HasNorms() ? "norms" : "no norms") + "; " + (mergeState.FieldInfos.HasDocValues() ? "docValues" : "no docValues") + "; " + (mergeState.FieldInfos.HasProx() ? "prox" : "no prox") + "; " + (mergeState.FieldInfos.HasProx() ? "freqs" : "no freqs"));
}
}
// 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);
bool useCompoundFile;
lock (this) // Guard segmentInfos
{
useCompoundFile = mergePolicy.UseCompoundFile(segmentInfos, merge.Info_Renamed);
}
if (useCompoundFile)
{
success = false;
ICollection<string> filesToRemove = merge.Info_Renamed.Files();
try
{
filesToRemove = CreateCompoundFile(infoStream, directory, checkAbort, merge.Info_Renamed.Info, context);
success = true;
}
catch (System.IO.IOException ioe)
{
lock (this)
{
if (merge.Aborted)
{
// this can happen if rollback or close(false)
// is called -- fall through to logic below to
// remove the partially created CFS:
}
else
{
HandleMergeException(ioe, merge);
}
}
}
catch (Exception t)
{
HandleMergeException(t, merge);
}
finally
{
if (!success)
{
if (infoStream.IsEnabled("IW"))
{
infoStream.Message("IW", "hit exception creating compound file during merge");
}
lock (this)
{
Deleter.DeleteFile(Lucene.Net.Index.IndexFileNames.SegmentFileName(mergedName, "", Lucene.Net.Index.IndexFileNames.COMPOUND_FILE_EXTENSION));
Deleter.DeleteFile(Lucene.Net.Index.IndexFileNames.SegmentFileName(mergedName, "", Lucene.Net.Index.IndexFileNames.COMPOUND_FILE_ENTRIES_EXTENSION));
Deleter.DeleteNewFiles(merge.Info_Renamed.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;
lock (this)
{
// delete new non cfs files directly: they were never
// registered with IFD
Deleter.DeleteNewFiles(filesToRemove);
if (merge.Aborted)
{
if (infoStream.IsEnabled("IW"))
{
infoStream.Message("IW", "abort merge after building CFS");
}
Deleter.DeleteFile(Lucene.Net.Index.IndexFileNames.SegmentFileName(mergedName, "", Lucene.Net.Index.IndexFileNames.COMPOUND_FILE_EXTENSION));
Deleter.DeleteFile(Lucene.Net.Index.IndexFileNames.SegmentFileName(mergedName, "", Lucene.Net.Index.IndexFileNames.COMPOUND_FILE_ENTRIES_EXTENSION));
return 0;
}
}
merge.Info_Renamed.Info.UseCompoundFile = 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:
bool success2 = false;
try
{
Codec.SegmentInfoFormat().SegmentInfoWriter.Write(directory, merge.Info_Renamed.Info, mergeState.FieldInfos, context);
success2 = true;
}
finally
{
if (!success2)
{
lock (this)
{
Deleter.DeleteNewFiles(merge.Info_Renamed.Files());
}
}
}
// TODO: ideally we would freeze merge.info here!!
// because any changes after writing the .si will be
// lost...
if (infoStream.IsEnabled("IW"))
{
infoStream.Message("IW", string.Format(CultureInfo.InvariantCulture, "merged segment size=%.3f MB vs estimate=%.3f MB", merge.Info_Renamed.SizeInBytes() / 1024.0 / 1024.0, merge.EstimatedMergeBytes / 1024 / 1024.0));
}
IndexReaderWarmer mergedSegmentWarmer = Config_Renamed.MergedSegmentWarmer;
if (PoolReaders && mergedSegmentWarmer != null && merge.Info_Renamed.Info.DocCount != 0)
{
ReadersAndUpdates rld = readerPool.Get(merge.Info_Renamed, true);
SegmentReader sr = rld.GetReader(IOContext.READ);
try
{
mergedSegmentWarmer.Warm(sr);
}
finally
{
lock (this)
{
rld.Release(sr);
readerPool.Release(rld);
}
}
}
// Force READ context because we merge deletes onto
// this reader:
if (!CommitMerge(merge, mergeState))
{
// commitMerge will return false if this merge was
// aborted
return 0;
}
success = true;
}
finally
{
// Readers are already closed in commitMerge if we didn't hit
// an exc:
if (!success)
{
CloseMergeReaders(merge, true);
}
}
return merge.Info_Renamed.Info.DocCount;
}
/// <summary>Does the actual (time-consuming) work of the merge,
/// but without holding synchronized lock on IndexWriter
/// instance
/// </summary>
private int MergeMiddle(MergePolicy.OneMerge merge)
{
merge.CheckAborted(directory);
System.String mergedName = merge.info.name;
SegmentMerger merger = null;
int mergedDocCount = 0;
SegmentInfos sourceSegments = merge.segments;
int numSegments = sourceSegments.Count;
if (infoStream != null)
Message("merging " + merge.SegString(directory));
merger = new SegmentMerger(this, mergedName, merge);
merge.readers = new SegmentReader[numSegments];
merge.readersClone = new SegmentReader[numSegments];
bool mergeDocStores = false;
System.Collections.Hashtable dss = new System.Collections.Hashtable();
String currentDocStoreSegment;
lock(this) {
currentDocStoreSegment = docWriter.GetDocStoreSegment();
}
bool currentDSSMerged = false;
// This is try/finally to make sure merger's readers are
// closed:
bool success = false;
try
{
int totDocCount = 0;
for (int i = 0; i < numSegments; i++)
{
SegmentInfo info = sourceSegments.Info(i);
// Hold onto the "live" reader; we will use this to
// commit merged deletes
SegmentReader reader = merge.readers[i] = readerPool.Get(info, merge.mergeDocStores, MERGE_READ_BUFFER_SIZE, -1);
// We clone the segment readers because other
// deletes may come in while we're merging so we
// need readers that will not change
SegmentReader clone = merge.readersClone[i] = (SegmentReader)reader.Clone(true);
merger.Add(clone);
if (clone.HasDeletions())
{
mergeDocStores = true;
}
if (info.GetDocStoreOffset() != -1 && currentDocStoreSegment != null)
{
currentDSSMerged |= currentDocStoreSegment.Equals(info.GetDocStoreSegment());
}
totDocCount += clone.NumDocs();
}
if (infoStream != null)
{
Message("merge: total " + totDocCount + " docs");
}
merge.CheckAborted(directory);
// If deletions have arrived and it has now become
// necessary to merge doc stores, go and open them:
if (mergeDocStores && !merge.mergeDocStores)
{
merge.mergeDocStores = true;
lock (this)
{
if (currentDSSMerged)
{
if (infoStream != null)
{
Message("now flush at mergeMiddle");
}
DoFlush(true, false);
}
}
for (int i = 0; i < numSegments; i++)
{
merge.readersClone[i].OpenDocStores();
}
// Clear DSS
merge.info.SetDocStore(-1, null, false);
}
// This is where all the work happens:
mergedDocCount = merge.info.docCount = merger.Merge(merge.mergeDocStores);
System.Diagnostics.Debug.Assert(mergedDocCount == totDocCount);
if (merge.useCompoundFile)
{
success = false;
string compoundFileName = IndexFileNames.SegmentFileName(mergedName, IndexFileNames.COMPOUND_FILE_EXTENSION);
try
{
if (infoStream != null)
{
Message("create compound file " + compoundFileName);
}
merger.CreateCompoundFile(compoundFileName);
success = true;
}
catch (System.IO.IOException ioe)
{
lock (this)
{
if (merge.IsAborted())
{
// This can happen if rollback or close(false)
// is called -- fall through to logic below to
// remove the partially created CFS:
}
else
{
HandleMergeException(ioe, merge);
}
}
}
catch (Exception t)
{
HandleMergeException(t, merge);
}
finally
{
if (!success)
{
if (infoStream != null)
{
Message("hit exception creating compound file during merge");
}
lock (this)
{
deleter.DeleteFile(compoundFileName);
deleter.DeleteNewFiles(merger.GetMergedFiles());
}
}
}
success = false;
lock (this)
{
// delete new non cfs files directly: they were never
// registered with IFD
deleter.DeleteNewFiles(merger.GetMergedFiles());
if (merge.IsAborted())
{
if (infoStream != null)
{
Message("abort merge after building CFS");
}
deleter.DeleteFile(compoundFileName);
return 0;
}
}
merge.info.SetUseCompoundFile(true);
}
int termsIndexDivisor;
bool loadDocStores;
// if the merged segment warmer was not installed when
// this merge was started, causing us to not force
// the docStores to close, we can't warm it now
bool canWarm = merge.info.GetDocStoreSegment() == null || currentDocStoreSegment == null || !merge.info.GetDocStoreSegment().Equals(currentDocStoreSegment);
if (poolReaders && mergedSegmentWarmer != null && canWarm)
{
// Load terms index & doc stores so the segment
// warmer can run searches, load documents/term
// vectors
termsIndexDivisor = readerTermsIndexDivisor;
loadDocStores = true;
}
else
{
termsIndexDivisor = -1;
loadDocStores = false;
}
// TODO: in the non-realtime case, we may want to only
// keep deletes (it's costly to open entire reader
// when we just need deletes)
SegmentReader mergedReader = readerPool.Get(merge.info, loadDocStores, BufferedIndexInput.BUFFER_SIZE, termsIndexDivisor);
try
{
if (poolReaders && mergedSegmentWarmer != null)
{
mergedSegmentWarmer.Warm(mergedReader);
}
if (!CommitMerge(merge, merger, mergedDocCount, mergedReader))
{
// commitMerge will return false if this merge was aborted
return 0;
}
}
finally
{
lock (this)
{
readerPool.Release(mergedReader);
}
}
success = true;
}
finally
{
// Readers are already closed in commitMerge if we didn't hit
// an exc:
if (!success)
{
CloseMergeReaders(merge, true);
}
}
merge.mergeDone = true;
lock (mergeScheduler)
{
System.Threading.Monitor.PulseAll(mergeScheduler);
}
// Force a sync after commiting the merge. Once this
// sync completes then all index files referenced by the
// current segmentInfos are on stable storage so if the
// OS/machine crashes, or power cord is yanked, the
// index will be intact. Note that this is just one
// (somewhat arbitrary) policy; we could try other
// policies like only sync if it's been > X minutes or
// more than Y bytes have been written, etc.
if (autoCommit)
{
long size;
lock (this)
{
size = merge.info.SizeInBytes();
}
Commit(size);
}
return mergedDocCount;
}
/** Does the actual (time-consuming) work of the merge,
* but without holding synchronized lock on IndexWriter
* instance */
private int MergeMiddle(MergePolicy.OneMerge merge)
{
merge.CheckAborted(directory);
string mergedName = merge.info.name;
SegmentMerger merger = null;
int mergedDocCount = 0;
SegmentInfos sourceSegments = merge.segments;
SegmentInfos sourceSegmentsClone = merge.segmentsClone;
int numSegments = sourceSegments.Count;
if (infoStream != null)
Message("merging " + merge.SegString(directory));
merger = new SegmentMerger(this, mergedName, merge);
bool success = false;
// This is try/finally to make sure merger's readers are
// closed:
try
{
int totDocCount = 0;
for (int i = 0; i < numSegments; i++)
{
SegmentInfo si = sourceSegmentsClone.Info(i);
IndexReader reader = SegmentReader.Get(true, si, MERGE_READ_BUFFER_SIZE, merge.mergeDocStores); // no need to set deleter (yet)
merger.Add(reader);
totDocCount += reader.NumDocs();
}
if (infoStream != null)
{
Message("merge: total " + totDocCount + " docs");
}
merge.CheckAborted(directory);
// This is where all the work happens:
mergedDocCount = merge.info.docCount = merger.Merge(merge.mergeDocStores);
System.Diagnostics.Debug.Assert(mergedDocCount == totDocCount);
success = true;
}
finally
{
// close readers before we attempt to delete
// now-obsolete segments
if (merger != null)
{
merger.CloseReaders();
}
}
if (!CommitMerge(merge, merger, mergedDocCount))
// commitMerge will return false if this merge was aborted
return 0;
if (merge.useCompoundFile)
{
// Maybe force a sync here to allow reclaiming of the
// disk space used by the segments we just merged:
if (autoCommit && DoCommitBeforeMergeCFS(merge))
{
long size;
lock (this)
{
size = merge.info.SizeInBytes();
}
Commit(size);
}
success = false;
string compoundFileName = mergedName + "." + IndexFileNames.COMPOUND_FILE_EXTENSION;
try
{
merger.CreateCompoundFile(compoundFileName);
success = true;
}
catch (System.IO.IOException ioe)
{
lock (this)
{
if (merge.IsAborted())
{
// This can happen if rollback or close(false)
// is called -- fall through to logic below to
// remove the partially created CFS:
success = true;
}
else
HandleMergeException(ioe, merge);
}
}
catch (System.Exception t)
{
HandleMergeException(t, merge);
}
finally
{
if (!success)
{
if (infoStream != null)
Message("hit exception creating compound file during merge");
lock (this)
{
deleter.DeleteFile(compoundFileName);
}
}
}
if (merge.IsAborted())
{
if (infoStream != null)
Message("abort merge after building CFS");
deleter.DeleteFile(compoundFileName);
return 0;
}
lock (this)
{
if (segmentInfos.IndexOf(merge.info) == -1 || merge.IsAborted())
{
// Our segment (committed in non-compound
// format) got merged away while we were
// building the compound format.
deleter.DeleteFile(compoundFileName);
}
else
{
merge.info.SetUseCompoundFile(true);
Checkpoint();
}
}
}
// Force a sync after commiting the merge. Once this
// sync completes then all index files referenced by the
// current segmentInfos are on stable storage so if the
// OS/machine crashes, or power cord is yanked, the
// index will be intact. Note that this is just one
// (somewhat arbitrary) policy; we could try other
// policies like only sync if it's been > X minutes or
// more than Y bytes have been written, etc.
if (autoCommit)
{
long size;
lock (this)
{
size = merge.info.SizeInBytes();
}
Commit(size);
}
return mergedDocCount;
}
/// <summary>Does the actual (time-consuming) work of the merge,
/// but without holding synchronized lock on IndexWriter
/// instance
/// </summary>
private int MergeMiddle(MergePolicy.OneMerge merge)
{
merge.CheckAborted(directory);
System.String mergedName = merge.info.name;
SegmentMerger merger = null;
int mergedDocCount = 0;
SegmentInfos sourceSegments = merge.segments;
int numSegments = sourceSegments.Count;
if (infoStream != null)
Message("merging " + merge.SegString(directory));
merger = new SegmentMerger(this, mergedName, merge);
merge.readers = new SegmentReader[numSegments];
merge.readersClone = new SegmentReader[numSegments];
bool mergeDocStores = false;
System.Collections.Hashtable dss = new System.Collections.Hashtable();
// This is try/finally to make sure merger's readers are
// closed:
bool success = false;
try
{
int totDocCount = 0;
for (int i = 0; i < numSegments; i++)
{
SegmentInfo info = sourceSegments.Info(i);
// Hold onto the "live" reader; we will use this to
// commit merged deletes
SegmentReader reader = merge.readers[i] = readerPool.Get(info, merge.mergeDocStores, MERGE_READ_BUFFER_SIZE, - 1);
// We clone the segment readers because other
// deletes may come in while we're merging so we
// need readers that will not change
SegmentReader clone = merge.readersClone[i] = (SegmentReader) reader.Clone(true);
merger.Add(clone);
if (clone.HasDeletions())
{
mergeDocStores = true;
}
if (info.GetDocStoreOffset() != - 1)
{
dss[info.GetDocStoreSegment()] = info.GetDocStoreSegment();
}
totDocCount += clone.NumDocs();
}
if (infoStream != null)
{
Message("merge: total " + totDocCount + " docs");
}
merge.CheckAborted(directory);
// If deletions have arrived and it has now become
// necessary to merge doc stores, go and open them:
if (mergeDocStores && !merge.mergeDocStores)
{
merge.mergeDocStores = true;
lock (this)
{
String key = docWriter.GetDocStoreSegment();
if (key!=null && dss.Contains(key))
{
if (infoStream != null)
Message("now flush at mergeMiddle");
DoFlush(true, false);
}
}
for (int i = 0; i < numSegments; i++)
{
merge.readersClone[i].OpenDocStores();
}
// Clear DSS
lock (this)
{
merge.info.SetDocStore(- 1, null, false);
}
}
// This is where all the work happens:
mergedDocCount = merge.info.docCount = merger.Merge(merge.mergeDocStores);
System.Diagnostics.Debug.Assert(mergedDocCount == totDocCount);
// TODO: in the non-realtime case, we may want to only
// keep deletes (it's costly to open entire reader
// when we just need deletes)
SegmentReader mergedReader = readerPool.Get(merge.info, false, BufferedIndexInput.BUFFER_SIZE, - 1);
try
{
if (poolReaders && mergedSegmentWarmer != null)
{
mergedSegmentWarmer.Warm(mergedReader);
}
if (!CommitMerge(merge, merger, mergedDocCount, mergedReader))
// commitMerge will return false if this merge was aborted
return 0;
}
finally
{
lock (this)
{
readerPool.Release(mergedReader);
}
}
success = true;
}
finally
{
lock (this)
{
if (!success)
{
// Suppress any new exceptions so we throw the
// original cause
for (int i = 0; i < numSegments; i++)
{
if (merge.readers[i] != null)
{
try
{
readerPool.Release(merge.readers[i], true);
}
catch (System.Exception t)
{
}
}
if (merge.readersClone[i] != null)
{
try
{
merge.readersClone[i].Close();
}
catch (System.Exception t)
{
}
// This was a private clone and we had the only reference
System.Diagnostics.Debug.Assert(merge.readersClone[i].GetRefCount() == 0);
}
}
}
else
{
for (int i = 0; i < numSegments; i++)
{
if (merge.readers[i] != null)
{
readerPool.Release(merge.readers[i], true);
}
if (merge.readersClone[i] != null)
{
merge.readersClone[i].Close();
// This was a private clone and we had the only reference
System.Diagnostics.Debug.Assert(merge.readersClone[i].GetRefCount() == 0);
}
}
}
}
}
// Must checkpoint before decrefing so any newly
// referenced files in the new merge.info are incref'd
// first:
lock (this)
{
deleter.Checkpoint(segmentInfos, false);
}
DecrefMergeSegments(merge);
if (merge.useCompoundFile)
{
// Maybe force a sync here to allow reclaiming of the
// disk space used by the segments we just merged:
if (autoCommit && DoCommitBeforeMergeCFS(merge))
{
long size;
lock (this)
{
size = merge.info.SizeInBytes();
}
Commit(size);
}
success = false;
System.String compoundFileName = mergedName + "." + IndexFileNames.COMPOUND_FILE_EXTENSION;
try
{
merger.CreateCompoundFile(compoundFileName);
success = true;
}
catch (System.IO.IOException ioe)
{
lock (this)
{
if (merge.IsAborted())
{
// This can happen if rollback or close(false)
// is called -- fall through to logic below to
// remove the partially created CFS:
success = true;
}
else
HandleMergeException(ioe, merge);
}
}
catch (System.Exception t)
{
HandleMergeException(t, merge);
}
finally
{
if (!success)
{
if (infoStream != null)
Message("hit exception creating compound file during merge");
lock (this)
{
deleter.DeleteFile(compoundFileName);
}
}
}
if (merge.IsAborted())
{
if (infoStream != null)
Message("abort merge after building CFS");
deleter.DeleteFile(compoundFileName);
return 0;
}
lock (this)
{
if (segmentInfos.IndexOf(merge.info) == - 1 || merge.IsAborted())
{
// Our segment (committed in non-compound
// format) got merged away while we were
// building the compound format.
deleter.DeleteFile(compoundFileName);
}
else
{
merge.info.SetUseCompoundFile(true);
Checkpoint();
}
}
}
// Force a sync after commiting the merge. Once this
// sync completes then all index files referenced by the
// current segmentInfos are on stable storage so if the
// OS/machine crashes, or power cord is yanked, the
// index will be intact. Note that this is just one
// (somewhat arbitrary) policy; we could try other
// policies like only sync if it's been > X minutes or
// more than Y bytes have been written, etc.
if (autoCommit)
{
long size;
lock (this)
{
size = merge.info.SizeInBytes();
}
Commit(size);
}
return mergedDocCount;
}
/// <summary>Does the actual (time-consuming) work of the merge,
/// but without holding synchronized lock on IndexWriter
/// instance
/// </summary>
private int MergeMiddle(MergePolicy.OneMerge merge)
{
merge.CheckAborted(directory);
System.String mergedName = merge.info.name;
SegmentMerger merger = null;
int mergedDocCount = 0;
SegmentInfos sourceSegments = merge.segments;
SegmentInfos sourceSegmentsClone = merge.segmentsClone;
int numSegments = sourceSegments.Count;
if (infoStream != null)
Message("merging " + merge.SegString(directory));
merger = new SegmentMerger(this, mergedName, merge);
// This is try/finally to make sure merger's readers are
// closed:
bool success = false;
try
{
int totDocCount = 0;
for (int i = 0; i < numSegments; i++)
{
SegmentInfo si = sourceSegmentsClone.Info(i);
IndexReader reader = SegmentReader.Get(si, MERGE_READ_BUFFER_SIZE, merge.mergeDocStores); // no need to set deleter (yet)
merger.Add(reader);
totDocCount += reader.NumDocs();
}
if (infoStream != null)
{
Message("merge: total " + totDocCount + " docs");
}
merge.CheckAborted(directory);
mergedDocCount = merge.info.docCount = merger.Merge(merge.mergeDocStores);
System.Diagnostics.Debug.Assert(mergedDocCount == totDocCount);
success = true;
}
finally
{
// close readers before we attempt to delete
// now-obsolete segments
if (merger != null)
{
merger.CloseReaders();
}
if (!success)
{
if (infoStream != null)
Message("hit exception during merge; now refresh deleter on segment " + mergedName);
lock (this)
{
AddMergeException(merge);
deleter.Refresh(mergedName);
}
}
}
if (!CommitMerge(merge))
// commitMerge will return false if this merge was aborted
return 0;
if (merge.useCompoundFile)
{
success = false;
bool skip = false;
System.String compoundFileName = mergedName + "." + IndexFileNames.COMPOUND_FILE_EXTENSION;
try
{
try
{
merger.CreateCompoundFile(compoundFileName);
success = true;
}
catch (System.IO.IOException ioe)
{
lock (this)
{
if (segmentInfos.IndexOf(merge.info) == - 1)
{
// If another merge kicked in and merged our
// new segment away while we were trying to
// build the compound file, we can hit a
// FileNotFoundException and possibly
// IOException over NFS. We can tell this has
// happened because our SegmentInfo is no
// longer in the segments; if this has
// happened it is safe to ignore the exception
// & skip finishing/committing our compound
// file creating.
if (infoStream != null)
Message("hit exception creating compound file; ignoring it because our info (segment " + merge.info.name + ") has been merged away");
skip = true;
}
else
throw ioe;
}
}
}
finally
{
if (!success)
{
if (infoStream != null)
Message("hit exception creating compound file during merge: skip=" + skip);
lock (this)
{
if (!skip)
AddMergeException(merge);
deleter.DeleteFile(compoundFileName);
}
}
}
if (!skip)
{
lock (this)
{
if (skip || segmentInfos.IndexOf(merge.info) == - 1 || merge.IsAborted())
{
// Our segment (committed in non-compound
// format) got merged away while we were
// building the compound format.
deleter.DeleteFile(compoundFileName);
}
else
{
success = false;
try
{
merge.info.SetUseCompoundFile(true);
Checkpoint();
success = true;
}
finally
{
if (!success)
{
if (infoStream != null)
Message("hit exception checkpointing compound file during merge");
// Must rollback:
AddMergeException(merge);
merge.info.SetUseCompoundFile(false);
DeletePartialSegmentsFile();
deleter.DeleteFile(compoundFileName);
}
}
// Give deleter a chance to remove files now.
deleter.Checkpoint(segmentInfos, autoCommit);
}
}
}
}
return mergedDocCount;
}
/* FIXME if we want to support non-contiguous segment merges */
private bool CommitMerge(MergePolicy.OneMerge merge)
{
lock (this)
{
System.Diagnostics.Debug.Assert(merge.registerDone);
if (hitOOM)
return false;
if (infoStream != null)
Message("CommitMerge: " + merge.SegString(directory));
// If merge was explicitly aborted, or, if abort() or
// rollbackTransaction() had been called since our merge
// started (which results in an unqualified
// deleter.refresh() call that will remove any index
// file that current segments does not reference), we
// abort this merge
if (merge.IsAborted())
{
if (infoStream != null)
Message("commitMerge: skipping merge " + merge.SegString(directory) + ": it was aborted");
System.Diagnostics.Debug.Assert(merge.increfDone);
DecrefMergeSegments(merge);
deleter.Refresh(merge.info.name);
return false;
}
bool success = false;
int start;
try
{
SegmentInfos sourceSegmentsClone = merge.segmentsClone;
SegmentInfos sourceSegments = merge.segments;
start = EnsureContiguousMerge(merge);
if (infoStream != null)
Message("commitMerge " + merge.SegString(directory));
// Carefully merge deletes that occurred after we
// started merging:
BitVector deletes = null;
int docUpto = 0;
int numSegmentsToMerge = sourceSegments.Count;
for (int i = 0; i < numSegmentsToMerge; i++)
{
SegmentInfo previousInfo = sourceSegmentsClone.Info(i);
SegmentInfo currentInfo = sourceSegments.Info(i);
System.Diagnostics.Debug.Assert(currentInfo.docCount == previousInfo.docCount);
int docCount = currentInfo.docCount;
if (previousInfo.HasDeletions())
{
// There were deletes on this segment when the merge
// started. The merge has collapsed away those
// deletes, but, if new deletes were flushed since
// the merge started, we must now carefully keep any
// newly flushed deletes but mapping them to the new
// docIDs.
System.Diagnostics.Debug.Assert(currentInfo.HasDeletions());
// Load deletes present @ start of merge, for this segment:
BitVector previousDeletes = new BitVector(previousInfo.dir, previousInfo.GetDelFileName());
if (!currentInfo.GetDelFileName().Equals(previousInfo.GetDelFileName()))
{
// This means this segment has had new deletes
// committed since we started the merge, so we
// must merge them:
if (deletes == null)
deletes = new BitVector(merge.info.docCount);
BitVector currentDeletes = new BitVector(currentInfo.dir, currentInfo.GetDelFileName());
for (int j = 0; j < docCount; j++)
{
if (previousDeletes.Get(j))
System.Diagnostics.Debug.Assert(currentDeletes.Get(j));
else
{
if (currentDeletes.Get(j))
deletes.Set(docUpto);
docUpto++;
}
}
}
else
docUpto += docCount - previousDeletes.Count();
}
else if (currentInfo.HasDeletions())
{
// This segment had no deletes before but now it
// does:
if (deletes == null)
deletes = new BitVector(merge.info.docCount);
BitVector currentDeletes = new BitVector(directory, currentInfo.GetDelFileName());
for (int j = 0; j < docCount; j++)
{
if (currentDeletes.Get(j))
deletes.Set(docUpto);
docUpto++;
}
}
// No deletes before or after
else
docUpto += currentInfo.docCount;
merge.CheckAborted(directory);
}
if (deletes != null)
{
merge.info.AdvanceDelGen();
deletes.Write(directory, merge.info.GetDelFileName());
}
success = true;
}
finally
{
if (!success)
{
if (infoStream != null)
Message("hit exception creating merged deletes file");
deleter.Refresh(merge.info.name);
}
}
// Simple optimization: if the doc store we are using
// has been closed and is in now compound format (but
// wasn't when we started), then we will switch to the
// compound format as well:
System.String mergeDocStoreSegment = merge.info.GetDocStoreSegment();
if (mergeDocStoreSegment != null && !merge.info.GetDocStoreIsCompoundFile())
{
int size = segmentInfos.Count;
for (int i = 0; i < size; i++)
{
SegmentInfo info = segmentInfos.Info(i);
System.String docStoreSegment = info.GetDocStoreSegment();
if (docStoreSegment != null && docStoreSegment.Equals(mergeDocStoreSegment) && info.GetDocStoreIsCompoundFile())
{
merge.info.SetDocStoreIsCompoundFile(true);
break;
}
}
}
success = false;
SegmentInfos rollback = null;
try
{
rollback = (SegmentInfos) segmentInfos.Clone();
((System.Collections.IList) ((System.Collections.ArrayList) segmentInfos).GetRange(start, start + merge.segments.Count - start)).Clear();
segmentInfos.Insert(start, merge.info);
Checkpoint();
success = true;
}
finally
{
if (!success && rollback != null)
{
if (infoStream != null)
Message("hit exception when checkpointing after merge");
segmentInfos.Clear();
segmentInfos.AddRange(rollback);
DeletePartialSegmentsFile();
deleter.Refresh(merge.info.name);
}
}
if (merge.optimize)
segmentsToOptimize.Add(merge.info, merge.info);
// Must checkpoint before decrefing so any newly
// referenced files in the new merge.info are incref'd
// first:
deleter.Checkpoint(segmentInfos, autoCommit);
DecrefMergeSegments(merge);
return true;
}
}
