On 12/14/2015 10:47 PM, ig01 wrote: > We installed solr with solr.cmd -e cloud utility that comes with the > installation. > The names of shards are odd because in this case after the installation > We've migrated an old index from our other environment (wich is solr single > node) and splitted it with Collection API splitt command. > The splitting completed successfuly, documents were spreaded almost equally > between two shards and I was able to retrieve our old documents. After that > I deleted the old shard that was splitted (with Collection API delete > command). > > Anyway this behavior is the same also for a regular solr cloud installation > with solr.cmd -e cloud, without any index migration...
The solr.cmd script did not exist in Solr 4.4, which your initial message on this thread indicated you were running. If we ignore that and assume you're running a 5.x version, then we reach another snag. The "-e cloud" option is not for installation -- that's for setting up a multi-node SolrCloud *example* with all nodes running on the same host and one zookeeper node embedded in the first Solr node. A production installation of SolrCloud should have multiple servers. Each Solr server will most likely be running one SolrCloud node, and three or five of your servers will be running a standalone zookeeper, part of a zookeeper ensemble. > We are indexing our documents by using the > url="http://10.1.20.31/8983/solr/collection1/". > After the installation we indexed 40000 documents and they all indexed on > the same shard. This doesn't give any information about how you are indexing. This is a variable assignment, telling your program where to find Solr. Your indexing program could be using any of dozens of Solr libraries, or a program that uses pure HTTP, constructing URLs internally. It could even be a shell script using curl. Thanks, Shawn