Hmm, im not sure, i don't use this kind of crawling but i can imagine input dir segments/* does not exist. Try to remove the asterisk? If that doesn't work, how much free disk space you have in your tmp directory? Then try setting hadoop.tmp.dir to a disk with plenty of room.
> 2010-11-04 13:48:00,555 WARN segment.SegmentMerger - Input dir > /lib/nutch/crawl/segments/* doesn't exist, skipping.
