It looks like CouchDB is not the only NoSQL store that uses Lucene for full text indexing. Neo4j, the graph database, has no built-in indexing features, but provides a plugable mechanism for supporting it. You can read more about this integration on ☞ Neo4j wiki.
There is also a post from Arin Sarkissian providing ☞ a quick example of how node indexing should be implemented.
While I do appreciate the fact that these projects are not suffering from the “not invented here” syndrome (and I read that Lucene can scale), I would definitely find very useful to see some good references/recommendations on how to deal with Lucene scaling once Lucene-based full text/node indexing is used.
Update: Neo4j is getting closer to its 1.0 release and the latest RCs include some improvements on the node indexing. You can read more about it in the ☞ changelog