elasticsearch-node is a Ruby library to start, manipulate and shutdown elasticsearch nodes in a controlled fashion. On JRuby, it also provides a native binding to this functionality. Its very handy in testing environments.

gem install elasticsearch-node

After installation, make sure to run:

install_elasticsearch

For legacy versions, use:

ES_VERSION=0.18.7

require 'elasticsearch-node/external'
require 'elasticsearch-node/embedded'
require 'elasticsearch-node/remote'

external = ElasticSearch::Node::External.new "gateway.type" => "none"
embedded = ElasticSearch::Node::Embedded.new "gateway.type" => "none"
remote   = ElasticSearch::Node::Remote.new '127.0.0.1', 9200

[external, embedded, remote].each do |node|
  puts node.ip
  puts node.port
end

embedded.client

elasticsearch-node provides 3 types of nodes: External, Embedded and Remote. They differ in capabilities:

• External nodes are started as an external process and can be started, configured and shut down
• Embedded nodes are only available on JRuby and start and embedded Elasticsearch instance. In addition to the capabilities of external nodes, embedded nodes expose the client method, which returns a vanilla Elasticsearch client
• Remote nodes a standin for nodes not under control by this process: they are only configured using port and ip and have not further capabilities.

elasticsearch-node also provides a protocol to associate clients with nodes. See this example:

n = ElasticSearch::Node::External.new do
  def client
    conn = Faraday.new(:url => "http://#{self.ip}:#{self.port}") do |builder|
      builder.adapter :net_http
    end
  end
end

n.client #=> <#Faraday::Connection...

# or alternatively:

module HTTPClient
  def client
    conn = Faraday.new(:url => "http://#{self.ip}:#{self.port}") do |builder|
      builder.adapter :net_http
    end
  end
end

n = ElasticSearch::Node::External.new :client_module => HTTPClient

This allows client implementations to be independent of which kind of nodes they run on (as long as they don't rely on those features).

Please be aware that this library does not add any at_exit-hooks to your application. Shutdown must be taken care of, as elasticsearch takes some seconds to shut down and killing it can lead to data corruption. In most cases, its as easy as:

at_exit { node.close }

See COPYING.md for all details.