I'm not sure if Carrot does, but with Solr you have to specify which fields to cluster on then restart the instance. The clustering functionality is a contrib library I believe.