This happens when addIndexesNoOptimize finds a segment that's larger than maxMergeDocs in the index(es) it's given.

If you leave maxMergeDocs at Integer.MAX_VALUE it will fix that.

Though really it's being a little too pedantic because that setting (maxMergeDocs) sets the maximum size of a segment that's may be selected for merging. So it's easily possible to have segments in your index that are larger than this (eg by merging segments that were all just a bit smaller than the limit).

I think we should remove those checks and allow addIndexesNoOptimize to import and index even if it has segments over this limit. I'll open an issue.

Mike

vivek sar wrote:
Hi,

 I'm using 2.3.0 Lucene build and have following merge parameters,

mergeFactor = 100
maxMergeDocs = 99999
maxBufferedDocs = 10000
maxRAMBufferSizeMB = 200


After running with this setting for a month without problem  all the
sudden I'm getting following exception,

java.lang.IllegalArgumentException: Segment is too large (100001 docs
vs max docs 99999)
at org.apache.lucene.index.LogMergePolicy.findMerges (LogMergePolicy.java:274) at org.apache.lucene.index.IndexWriter.updatePendingMerges (IndexWriter.java:1824) at org.apache.lucene.index.IndexWriter.maybeMerge (IndexWriter.java:1799) at org.apache.lucene.index.IndexWriter.maybeMerge (IndexWriter.java:1795) at org.apache.lucene.index.IndexWriter.maybeMerge (IndexWriter.java:1791) at org.apache.lucene.index.IndexWriter.addIndexesNoOptimize (IndexWriter.java:2231)


I thought Lucene should control the number of documents in a segment
during merge. Is there a different configuration I need to set? Should
I leave the maxMergeDocs to its default value, Integer#MAX_VALUE?

Thanks,
-vivek

---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]



---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]

Reply via email to