I hope my letter will not be off-topic because I need general help/expirence with Lucene.
We provide hosting with our CMS in java/servlets. We have search feature there currently working via mnogosearch, but it poses a lot of limitations on configurability and generally suits our java environment poorly. We want to implement search using Lucene, and I need some suggestions from experienced users. First, what should I use? Should I take Lucene and some HTML document implementation (possibly from IndexHTML demo), or should I take Nutch? Will not Nutch be too hard on memory for case 50 sites running on one Resin container? How should I better index? On page add, or crawl the whole site at once each week? If I index one new document, how consuming is that, and do I instantly get ability to search throught it? Maybe I have to ask on nutch's list? But I'm unsure really whenever it's easy to integrate that, given its tools running one another classes' main().
