John Wang created LUCENE-6969:
---------------------------------

             Summary: Exception in reading SortedDocValues
                 Key: LUCENE-6969
                 URL: https://issues.apache.org/jira/browse/LUCENE-6969
             Project: Lucene - Core
          Issue Type: Bug
          Components: core/index
    Affects Versions: 5.4
            Reporter: John Wang


While reading SortedDocValues, I am getting the following stacktrace:
{noformat}
Exception in thread "main" java.lang.IndexOutOfBoundsException
        at java.nio.Buffer.checkIndex(Buffer.java:546)
        at java.nio.DirectByteBuffer.getShort(DirectByteBuffer.java:590)
        at 
org.apache.lucene.store.ByteBufferIndexInput$SingleBufferImpl.readShort(ByteBufferIndexInput.java:443)
        at 
org.apache.lucene.util.packed.DirectReader$DirectPackedReader16.get(DirectReader.java:185)
        at 
org.apache.lucene.codecs.lucene54.Lucene54DocValuesProducer$2.get(Lucene54DocValuesProducer.java:502)
        at org.apache.lucene.util.LongValues.get(LongValues.java:45)
        at 
org.apache.lucene.codecs.lucene54.Lucene54DocValuesProducer$7.getOrd(Lucene54DocValuesProducer.java:800)
{noformat}

I am not able to reproduce with a unit test. However, I am able to consistently 
reproduce it with my data. I think this is some sort of off-by-one error caused 
with my index.

I am happy to provide my index for reproduce this offline.

Here is the code snippet:
{noformat}
public static void main(String[] args) throws Exception {
    File idx = new File("/tmp/rapid/data/1448008627000_1448120881000");
    int id = 719265;
    Path idxPath = FileSystems.getDefault().getPath(idx.getAbsolutePath());
    FSDirectory dir = FSDirectory.open(idxPath);
    DirectoryReader dreader = DirectoryReader.open(dir);
    LeafReaderContext ctx = dreader.leaves().get(0);
        
    SortedDocValues docVal = ctx.reader().getSortedDocValues("host");
    int ord = docVal.getOrd(id);    
    System.out.println("ord: " + ord);
    
  }
{noformat}

My index has 1 segment and 5M docs.



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)

---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org
For additional commands, e-mail: dev-h...@lucene.apache.org

Reply via email to