A MultiReader is only limited by the Integer.MAX_VALUE limitation for the number of documents (because IndexReader.document() only uses int as doc-id). MultiReader has no logic at all, it is just a "container" of other index readers. The search runs on the leaves of the IndexReader (IndexReader.leaves()), which are AtomicReaders. The number of AtomicReaders on the whole reader structure (MultiReader with n DirectoryReaders each consisting of m AtomicReaders) and the number of documents per atomic reader is more the limiting factor.
----- Uwe Schindler H.-H.-Meier-Allee 63, D-28213 Bremen http://www.thetaphi.de eMail: u...@thetaphi.de > -----Original Message----- > From: Liz Sommers [mailto:lizswo...@gmail.com] > Sent: Friday, April 19, 2013 6:15 PM > To: java-user@lucene.apache.org > Subject: Is there a maximum size for a reader? > > Lucene 4.2.1 > > I am moving my application from 3.6 to 4.2.1. At the moment I have a reader > for each of my shards (100 shards each about 5 million records). I am > considering using a MultiReader for all the shards. The application gets > about > 2 million new records a week. Right now I don't have a full index to test on. > > My question is - approximately how many records can a Reader handle? > Does anybody know about memory constraints or size constraints? Do to > space constraints I can't build a full index for testing until I have > something > that I can nearly guarantee will work. > > Thanks > Liz --------------------------------------------------------------------- To unsubscribe, e-mail: java-user-unsubscr...@lucene.apache.org For additional commands, e-mail: java-user-h...@lucene.apache.org
