Requests to a node in your example would be answered by that node (no need to
distribute; it's a single shard system) and it would not internally be routed
otherwise either. Ultimately it is up to the client to load-balance the
initial requests into a "SolrCloud" cluster, but internally in a multi-shard
distributed search request it will be load balanced beyond that initial node.
CloudSolrServer does load balance, so if you're using that client it'll
"randomly" pick a shard to send to from the client-side. If you're using some
other mechanism, it'll request directly to whatever node that you've specified
directly for that initial request.
Erik
p.s. Thanks for attending the webinar, Bill! I saw your name as one of the
question askers. Hopefully all that stuff I made up is close to the truth :)
On Mar 27, 2013, at 14:51 , Bill Au wrote:
> I am running Solr 4.1. I have set up SolrCloud with 1 leader and 3
> replicas, 4 nodes total. Do query requests send to a node only query the
> replica on that node, or are they load-balanced to the entire cluster?
>
> Bill
