On 6/21/2013 3:32 AM, medley wrote: > I have this kind of url : > > "http://remoteserver/solr/leg0/select/?rows=10&version=2&fl=* > &shards= > remoteserver:80/solr/core0, > ... > remoteserver:80/solr/core5, > ... > remoteserver:80/solr/core9 > &.......... > > There is only ONE Solr instance with multiple cores (core0 to core9) on the > same server. > > I would like to know if it exists in Solr configuration files a mechanism > that makes the shard requests called locally on the server because it is the > same url and not to pass again through the (reverse) proxy in our case.
Building on what Upayavira already told you, don't use the name/address and port of the proxy in the shards request - use the names/addresses of the actual Solr servers with the actual Solr port. Also, you can put the shards parameter in the request handler definition, with one caveat - the core with the shards parameter cannot include itself on that shards parameter. If you do that, then it will try to include itself recursively and the request will never complete. Putting the shards parameter in the request handler definition is particularly powerful if you set up a core that has no index of its own, but has the same schema and a similar config, except for including the shards parameter and possibly the shards.qt parameter. I call this concept a broker core, but there is probably better terminology available. Thanks, Shawn
