Yes, I already had done this once, but is is not API conform any
more, when porting ndfs to hadoop is done I may can bring things
again to the api and provide a patch.
However there is a list of other issues on my todo list already so it
will not happen until next days.
Stefan
Am 04.02.2006 um 07:44 schrieb Chris Schneider:
Gang,
Would it be possible to modify Nutch so that a set of search
servers each had a local index, but that this index referred to
segments living in NDFS? Doing so would allow us to skip exporting
the segments from NDFS to the local FS. Of course, it would be
ideal to keep the crawling machines separate from the search
servers, but resource constraints do have a way of intruding... It
seems like this setup might provide sufficient performance, because
only the summaries in the current page of search results require
segment reading.
While we're on the subject of exporting segments to support
searching, which subdirs within the segment directories are
required for searching, building summaries, and displaying hits?
Thanks,
- Chris
--
------------------------
Chris Schneider
TransPac Software, Inc.
[EMAIL PROTECTED]
------------------------
