fix: free resources on close

Our transport class Manticore is explicitly threadsafe and by default creates
a pool with up to 50 connections; therefore we do not need an individual
instance of Manticore per worker thread. Using a shared instance resolves a
memory leak that was caused by using a strong reference to the worker threads
themselves as a key in the routing to per-thread clients, which prevented
those threads from being garbage-collected.

Additionally, Manticore gives us the opportunity to explicitly close the
client to free its resources in advance of garbage collection, so we should
propagate `LogStash::Filters::Base#close` to the shared client if it exists.

CHANGELOG.md

@@ -1,3 +1,6 @@
+## 3.15.3
+  - Fixes a memory leak that occurs when a pipeline containing this filter terminates, which could become significant if the pipeline is cycled repeatedly.
+
 ## 3.15.2
   - Added checking for `query` and `query_template`. [#171](https://github.com/logstash-plugins/logstash-filter-elasticsearch/pull/171)
 

lib/logstash/filters/elasticsearch.rb

@@ -4,6 +4,7 @@
 require "logstash/json"
 require 'logstash/plugin_mixins/ca_trusted_fingerprint_support'
 require "logstash/plugin_mixins/normalize_config_support"
+require "monitor"
 
 require_relative "elasticsearch/client"
 require_relative "elasticsearch/patches/_elasticsearch_transport_http_manticore"
@@ -139,7 +140,8 @@ class LogStash::Filters::Elasticsearch < LogStash::Filters::Base
 
   include LogStash::PluginMixins::NormalizeConfigSupport
 
-  attr_reader :clients_pool
+  include MonitorMixin
+  attr_reader :shared_client
 
   ##
   # @override to handle proxy => '' as if none was set
@@ -159,8 +161,6 @@ def self.validate_value(value, validator)
   end
 
   def register
-    @clients_pool = java.util.concurrent.ConcurrentHashMap.new
-
     #Load query if it exists
     if @query_template
       if File.zero?(@query_template)
@@ -244,6 +244,13 @@ def filter(event)
     end
   end # def filter
 
+  def close
+    synchronize do
+      @shared_client&.close
+    end
+    super
+  end
+
   # public only to be reuse in testing
   def prepare_user_agent
     os_name = java.lang.System.getProperty('os.name')
@@ -352,7 +359,9 @@ def new_client
   end
 
   def get_client
-    @clients_pool.computeIfAbsent(Thread.current, lambda { |x| new_client })
+    @shared_client || synchronize do
+      @shared_client ||= new_client
+    end
   end
 
   # get an array of path elements from a path reference

logstash-filter-elasticsearch.gemspec

@@ -1,7 +1,7 @@
 Gem::Specification.new do |s|
 
   s.name            = 'logstash-filter-elasticsearch'
-  s.version         = '3.15.2'
+  s.version         = '3.15.3'
   s.licenses        = ['Apache License (2.0)']
   s.summary         = "Copies fields from previous log events in Elasticsearch to current events "
   s.description     = "This gem is a Logstash plugin required to be installed on top of the Logstash core pipeline using $LS_HOME/bin/logstash-plugin install gemname. This gem is not a stand-alone program"

spec/filters/elasticsearch_spec.rb

@@ -91,21 +91,6 @@
       Thread.current[:filter_elasticsearch_client] = nil
     end
 
-    # Since the Elasticsearch Ruby client is not thread safe
-    # and under high load we can get error with the connection pool
-    # we have decided to create a new instance per worker thread which
-    # will be lazy created on the first call to `#filter`
-    #
-    # I am adding a simple test case for future changes
-    it "uses a different connection object per thread wait" do
-      expect(plugin.clients_pool.size).to eq(0)
-
-      Thread.new { plugin.filter(event) }.join
-      Thread.new { plugin.filter(event) }.join
-
-      expect(plugin.clients_pool.size).to eq(2)
-    end
-
     it "should enhance the current event with new data" do
       plugin.filter(event)
       expect(event.get("code")).to eq(404)
@@ -466,6 +451,24 @@ def wait_receive_request
       Thread.current[:filter_elasticsearch_client] = nil
     end
 
+    it 'uses a threadsafe transport adapter' do
+      client = plugin.send(:get_client).client
+      # we currently rely on the threadsafety guarantees provided by Manticore
+      # this spec is a safeguard to trigger an assessment of thread-safety should
+      # we choose a different transport adapter in the future.
+      expect(extract_transport(client).to_s).to include('Manticore')
+    end
+
+    it 'uses a single shared client across threads' do
+      q = Queue.new
+      10.times.map do
+        Thread.new(plugin) { |instance| q.push instance.send(:get_client) }
+      end.map(&:join)
+
+      first = q.pop
+      expect(q.pop).to be(first) until q.empty?
+    end
+
     describe "cloud.id" do
       let(:valid_cloud_id) do
         'sample:dXMtY2VudHJhbDEuZ2NwLmNsb3VkLmVzLmlvJGFjMzFlYmI5MDI0MTc3MzE1NzA0M2MzNGZkMjZmZDQ2OjkyNDMkYTRjMDYyMzBlNDhjOGZjZTdiZTg4YTA3NGEzYmIzZTA6OTI0NA=='