I did see that bug, which made me suspect Lucene. In my case, I tracked down
the problem. It was my own application. I was using Java's
FileChannel.transferTo functions to copy my index from one location to another.
One of the files is bigger than 2^31-1 bytes. So, one of my files was corrupted
during the copy because I was just doing one pass. I now loop the copy function
until the entire file is copied and everything works fine.
DOH!
----- Original Message ----
From: Yonik Seeley <[EMAIL PROTECTED]>
To: solr-user@lucene.apache.org
Sent: Wednesday, January 16, 2008 4:57:08 PM
Subject: Re: IOException: read past EOF during optimize phase
This may be a Lucene bug... IIRC, I saw at least one other lucene user
with a similar stack trace. I think the latest lucene version (2.3
dev) should fix it if that's the case.
-Yonik
On Jan 16, 2008 3:07 PM, Kevin Osborn <[EMAIL PROTECTED]> wrote:
> I am using the embedded Solr API for my indexing process. I created a
brand new index with my application without any problem. I then ran my
indexer in incremental mode. This process copies the working index to
a temporary Solr location, adds/updates any records, optimizes the
index, and then copies it back to the working location. There are currently
not any instances of Solr reading this index. Also, I commit after
every 100000 rows. The schema.xml and solrconfig.xml files have not
changed.
>
> Here is my function call.
> protected void optimizeProducts() throws IOException {
> UpdateHandler updateHandler = m_SolrCore.getUpdateHandler();
> CommitUpdateCommand commitCmd = new
CommitUpdateCommand(true);
> commitCmd.optimize = true;
>
> updateHandler.commit(commitCmd);
>
> log.info("Optimized index");
> }
>
> So, during the optimize phase, I get the following stack trace:
> java.io.IOException: read past EOF
> at
org.apache.lucene.store.BufferedIndexInput.refill(BufferedIndexInput.java:89)
> at
org.apache.lucene.store.BufferedIndexInput.readByte(BufferedIndexInput.java:34)
> at
org.apache.lucene.store.IndexInput.readChars(IndexInput.java:107)
> at
org.apache.lucene.store.IndexInput.readString(IndexInput.java:93)
> at
org.apache.lucene.index.FieldsReader.addFieldForMerge(FieldsReader.java:211)
> at
org.apache.lucene.index.FieldsReader.doc(FieldsReader.java:119)
> at
org.apache.lucene.index.SegmentReader.document(SegmentReader.java:323)
> at
org.apache.lucene.index.SegmentMerger.mergeFields(SegmentMerger.java:206)
> at
org.apache.lucene.index.SegmentMerger.merge(SegmentMerger.java:96)
> at
org.apache.lucene.index.IndexWriter.mergeSegments(IndexWriter.java:1835)
> at
org.apache.lucene.index.IndexWriter.optimize(IndexWriter.java:1195)
> at
org.apache.solr.update.DirectUpdateHandler2.commit(DirectUpdateHandler2.java:508)
> at ...
>
> There are no exceptions or anything else that appears to be incorrect
during the adds or commits. After this, the index files are still
non-optimized.
>
> I know there is not a whole lot to go on here. Anything in particular
that I should look at?
>
>
