CODEWITHSUNDEEP
lucenelucene.net
Pini Salim
Pini Salim

Reputation: 171

Lucene .Net optimization process

I am creating an index using Lucene .Net 2.9.2. After a lot of indexing, the index has many segments and deleted documents, so I am calling Optimize(numSegmets) on the IndexWriter.

The index's segments count is indeed reduced to the value of numSegmets, but it still has deletions... doesnt a call to Optimize should also remove all deleted documents?

My question is very important so I could know if this is how Lucene works or maybe I have some bug...

Edit: here is my code snippet:

IndexWriter writer = new IndexWriter(/*open writer from index directroy*/);
writer.Optimize(5);
writer.Commit();

bool hasDeletions = writer.HasDeletions();

hasDeletions is true, while I was expecting it would be false...

Upvotes: 2

Views: 2327

Answers (3)

Bill
Bill

Reputation: 11

Optimize seems to be deleting the entire index?

I am new to Lucene.NET - but I have it wired up and everything seems great! I added test data, removed items, and then tried to both optimize(1) and the ExpungeDeletes() (as seen above)...

but no matter how I approach this ... its not merging or whatever -- its just deleting the entire index?

my code looks like this (got it from a sample online):

public void Optimize()
{
    analyzer = new StandardAnalyzer(Version.LUCENE_30);
    using (var writer = new IndexWriter(_directory, analyzer, true, IndexWriter.MaxFieldLength.UNLIMITED))
    {
        analyzer.Close();
        //writer.Optimize(1);
        writer.ExpungeDeletes();
        writer.Dispose();
    }
}

I have no idea why this would delete the entire index?

Upvotes: 1

Christopher Currens
Christopher Currens

Reputation: 30695

Optimization merges segments, and during the segment merging, it removes the deletions that are listed in each one. If you don't do a full optimization, it's possible for deletions to remain, since the segments aren't merged/rebuilt.

This doesn't mean you need to do a full optimization in order to remove deletions.

IndexWriter writer = GetIndexWriter();
// delete some stuff
writer.ExpungeDeletes();

That will remove deleted documents from your index without doing a full optimization. It generally takes less time than an optimization, though it does depend on the MergePolicy, since it might still merge all the segments together (I believe by default it does not do this).

Upvotes: 2

jpountz
jpountz

Reputation: 9964

Deletions can remain unless you provide 1 as the maximum number of segments.

But you shouldn't worry about this. To quote the documentation for IndexWriter#optimize in Lucene 3.5

This method has been deprecated, as it is horribly inefficient and very rarely justified. Lucene's multi-segment search performance has improved over time, and the default TieredMergePolicy now targets segments with deletions.

Upvotes: 5

Related Questions