On 6/27/2014 7:47 AM, spirtt0 wrote: > I have some question about Solrcloud: > > 1. Is it possible to update dataimport configuration without full server > restart ? I was tried do that with core reload via API, but configuration > for dataimporthandler wasn't change
Are you updating a file on the disk, or changing the config in zookeeper? If you're updating a file on the disk, that doesn't do *anything* about the config files that are actually used, which are stored in zookeeper. Generally you'll need to upload a new config with zkcli. If you are using one of the bootstrap options, then a full restart of the servlet container might result in the new config getting uploaded ... but the bootstrap options are not the best way to maintain configurations. They are intended as a one-time option for converting from non-cloud to cloud. > 2. We want use SolrCloud with 3 replication nodes on separate servers. How > can i managed queries if one node is goes down ? Is it possible to have one > virtual IP that managed queries on new leader node - for HA cluster ? Solr doesn't manage IP addresses, that's up to you and your operating system. If you use the Java client (CloudSolrServer, part of SolrJ), it will always be aware of the cluster state and it will always work. If you're not using a cloud-aware client, then you need something like a load balancer in front of Solr. The software program called haproxy has been a good load balancer for my Solr install (not running SolrCloud). Thanks, Shawn
