Hello, I propose to join docs externally eg in tiny rdbms, just put ids there and keep content in files. Then DIH, I believe and only believe, should be able to build full document representation with joined entities.
As an alternative you can index document as is with id-references between them in separate solr core, then index joined docs into another core by DIH's SolrEntityProcessor querying the first core in with http://wiki.apache.org/solr/Join . 19.11.2012 23:55 пользователь "uwe72" <uwe.clem...@exxcellent.de<uwe.clem...@exxcellent.de>> написал: > Hi there, > > i have a principal question. > > We have arround 5 million lucene documents. > > At the beginning we have arround 4000 XML-files which we transform to > SolrInputDocuemnts by using solrj and adding them to the index. > > A document is also related to other documents, so while adding a document > we > have to do some queries (at least one) to identiy if there are related > documents already in the cache in order to do the association to the > related > document. The related document also has a "backlink", so we have to update > also the related document (means load, update, delete and re-add). > > We are using solr 3.6.1. > > The performance is quite slow because of this queries and modfifications of > already existing documents in the cache. > > Are there some configuration issues what we can do, or anything else? > > Thanks a lot in advance. > > > > > > -- > View this message in context: > http://lucene.472066.n3.nabble.com/Inserting-many-documents-and-update-relations-tp4021151.html > Sent from the Solr - User mailing list archive at Nabble.com. >
