Good afternoon (sorry for my english).

I'm a newbie for lucene. I'm working with Lucene 2.1.0 (core, analyzer and snowball) and with Alfresco Labs 3.0 (it's an ECM). During last months I noticed that the indexes were incomplete because I searched some documents and I had no results when I'm sure that this doc was uploaded successfully and the database shows me the information. I need to do a Full Reindex to recover it and solve this problem. This reindex spends about 10h. Once the recovery was completed I used Luke ToolBox to view the information and I noticed that its also incomplete (because I do searches and some documents were not viewed by lucene when they exists) and the index recovery log shows me that was successful at 100% completed.

I made another Full Reindex and when it was completed I noticed that I can find the documents that I couldn't see before but now I have other documents that I can't search but I can in the first full reindex. If I do another full reindex I will see some documents that I couldn't before but new documents will be lost and so on. I'm lost because I'm not able to find any coherency about why some documents are showed and why others not and vice versa if I do another reindex but not always the same documents.

Any suggestion will be appreciate. My lucene configuration is:

index.recovery.mode=AUTO
# index.recovery.mode=FULL
# Parametros anyadidos durante la instalacion
index.recovery.stopOnError=false
index.recovery.maximumPoolSize=5
index.tracking.cronExpression=0/5 * * * * ?
index.tracking.adm.cronExpression=${index.tracking .cronExpression}
index.tracking.avm.cronExpression=${index.tracking .cronExpression}
index.tracking.maxTxnDurationMinutes=10
index.tracking.reindexLagMs=10000
index.tracking.maxRecordSetSize=1000
index.tracking.maxTransactionsPerLuceneCommit=100
index.tracking.disableInTransactionIndexing=false
lucene.indexer.batchSize=1000
lucene.indexer.mergeFactor=10
lucene.query.maxClauses=10000
lucene.indexer.maxMergeDocs=100000
lucene.indexer.minMergeDocs=1000
#lucene.indexer.maxFieldLength=0 --> Don't index the content of the doc.

Many thanks.

Regards.