Indexing Issue between Mac OS X 10.5 and 10.6

[Kevin Murdoff <KM...@livescribe.com>]

Greetings Everyone -

I am hoping someone can help me with this unusual issue I have here.

Issue
Indexing information in a database (i.e.  /dataimport [full-import]) succeeds when I perform this function on a Mac OS X 10.6 with Java 1.6, but fails when I attempt the same indexing task on a 10.5 / Java 1.5 server.  When the indexing succeeds, I end up with 211,095 documents.  When the indexing fails (on the 10.5 machine), I end up with 58,286 documents.  The error I receive in the Tomcat 'catalina.out' log file is:

SEVERE: Full Import failed
org.apache.solr.handler.dataimport.DataImportHandlerException: java.lang.StackOverflowError
    at org.apache.solr.handler.dataimport.DocBuilder.buildDocument(DocBuilder.java:424)
    at org.apache.solr.handler.dataimport.DocBuilder.doFullDump(DocBuilder.java:242)
    at org.apache.solr.handler.dataimport.DocBuilder.execute(DocBuilder.java:180)
    at org.apache.solr.handler.dataimport.DataImporter.doFullImport(DataImporter.java:331)
    at org.apache.solr.handler.dataimport.DataImporter.runCmd(DataImporter.java:389)
    at org.apache.solr.handler.dataimport.DataImporter$1.run(DataImporter.java:370)
Caused by: java.lang.StackOverflowError
    at com.frontbase.jdbc.FBJRowHandler.close(Unknown Source)
    at com.frontbase.jdbc.FBJRowHandler.close(Unknown Source)
    ...

Background
I want to index the database information as a single document in Solr 1.4.1.  The document, as defined in the 'data-config.xml' file, has 10 entities, each with 5 primitive fields and 2 entity fields.  Most of these 10 entities do not represent very large datasets except one, which could represent over 95% of the result set.

I have tried tweaking the configuration values in the <mainIndex> section of the 'solrconfig.xml' file.  I lowered the <maxFieldLength> from 10,000 to 100, and lowered the <mergeFactor> from 10 to 5.  Making these changes, independently and together, did not exhibit any change in the indexing failures I have been experiencing.

I expanded the JVM min/max memory settings using -Xms and -Xmx set as high as 1024/2048 respectively.

I also obtained the Solr-1.4.1 release source code, built it on the 10.5 /1.5 server machine, and performed the same indexing task.  This resulted in the same stack overflow error.

Inquiry
Can someone tell me if they have experienced something similar?  If so, did you find a solution?  Or, does anyone know what may be causing these stack overflow errors?

Please let me know what other information I can provide that would be useful.

Thank you for your help!

- KFM