Hi Otis, thanks for the hint. Turns out I have 17.8 million unique terms. I'm fairly sure by now that the problem lies with the sorting. In the Lucene java docs (http://lucene.zones.apache.org:8080/hudson/job/Lucene-Nightly/javadoc/org/apache/lucene/search/Sort.html) it is stated that
> Sorting uses of caches of term values maintained by the internal HitQueue(s) Are these caches separate from the Solr caches? Is there a good way of reducing their max size when using Solr? Cheers, Chris Otis Gospodnetic wrote: > Chris, I checked Luke handler for you on a sample index. Indeed, it does > provide the number of terms and a bunch of other nice information, for > example: > > <int name="numDocs">19295605</int> > <int name="maxDoc">20437118</int> > <int name="numTerms">49209736</int> <------- here > <long name="version">1195333103547</long> > <bool name="optimized">false</bool> > <bool name="current">true</bool> > <bool name="hasDeletions">true</bool> > > > Otis > -- > Sematext -- http://sematext.com/ -- Lucene - Solr - Nutch >
