Yes, this should work. It will be very slow. There is a special hack by which you can say sort=_docid_+asc (or +desc). _docid_ is a magic field name that avoids sorting the results. Pulling documents at row # 1 million should be only a little slower than pulling documents at row #0.
On Mon, Mar 29, 2010 at 12:37 AM, Adrian Pemsel <apem...@gmail.com> wrote: > Hi, > > As part of our application I have written a reindex task that runs through > all documents in a core one by one (using *:*, a start offset and a row > limit of 1) and adds them to a new core (potentially with a new schema). > However, while working well for small sets this approach somehow does not > seem to work for larger data sets. The Reindex task counts its offset into > the old core, this count stops at about 118000 and no more documents are > returned. However, numDocs says there are around 582000 documents in the old > core. > Am I making a wrong assumption in believing I should get all documents like > this? > > Thanks, > > Adrian > -- Lance Norskog goks...@gmail.com
