Good idea. Thanks. Also, in such architecture (Master/Slave), is there any best practices for index stored on an NFS mounted filesystem ?
Specially about the rsync step, when the "slaves" want to synchronize their index from a remote filesystem (pb of inconsistent views of the directory). Nourredine. ________________________________ De : Noble Paul നോബിള് नोब्ळ् <noble.p...@corp.aol.com> À : solr-user@lucene.apache.org Envoyé le : Lundi, 14 Septembre 2009, 16h15mn 55s Objet : Re: Configuring "slaves" for a "master" backup without restarting you can put both master1 and master2 behind a VIP. If Master 1 goes down make the VIP point to Master2 On Mon, Sep 14, 2009 at 7:11 PM, nourredine khadri <nourredin...@yahoo.com> wrote: > Hi, > > A question about scalability. > > Let imagine the following architecture based on Master/Slave schema : > > - A "master" for the indexation called Master 1 > - A backup of Master 1 (called Master 2) > - Several "slaves" for search linked to Master 1 > > Can I configure the "slaves" to be automatically linked to Master 2 if Master > 1 fails without restarting the JVMs? > > Thanks in advance. > > Nourredine. > > > -- ----------------------------------------------------- Noble Paul | Principal Engineer| AOL | http://aol.com