Does anyone know what crawl output directories are required on a successful crawl? Are crawldb, indexes, index, linkdb and segments all required to have a successful merge?

I'm crawling on 5 servers and writing to the SAN. Everything goes fast and fine (up to several million documents). My problem is when I merge the indexes using the mergecrawls.sh, it takes a very long time. Is there any performance tuning that you can do to speed up the mergecrawls?

Thanks in advance,

Alex