All of the core loading stuff is on the server side, so CloudSolrServer isn't really germane (I don't think anyway).
This is in a bit of flux, so try having one core that's loaded on startup even if it's just a dummy core. There's currently ongoing work to play nicer with no cores being defined at startup, but that's not in 4.3. Take a look at: http://wiki.apache.org/solr/CoreAdmin#CREATE where it talks about optional parameters. NOTE: 4.4 (release imminent) has substantial fixes for the whole persistence situation. Also note that solr.xml is going away as a place to store core information and core discovery will be supported only from 5.x on. Good Luck! Erick On Mon, Jul 15, 2013 at 9:05 PM, Patrick Mi <patrick...@touchpointgroup.com> wrote: > Hi there, > > I run 2 solr instances ( Tomcat 7, Solr 4.3.0 , one shard),one external > Zookeeper instance and have lots of cores. > > I use collection API to create the new core dynamically after the > configuration for the core is uploaded to the Zookeeper and it all works > fine. > > As there are so many cores it takes very long time to load them at start up > I would like to start up the server quickly and load the cores on demand. > > When the core is created via collection API it is created with default > parameter : loadOnStartup="true" ( this can be seen in solr.xml ) > > Question: is there a way to specify this parameter so it can be set 'false' > in collection API ? > > Problem: If I manually set loadOnStartup="true" for the core I had exception > below when I used CloudSolrServer to query the core : > Error: org.apache.solr.client.solrj.SolrServerException: No live SolrServers > available to handle this request > > Seems to me that CloudSolrServer will not trigger the core to be loaded. > > Is it possible to get the core loaded using CloudSolrServer? > > Regards, > Patrick > >
