diff --git a/logstash-core/lib/logstash/java_pipeline.rb b/logstash-core/lib/logstash/java_pipeline.rb
index 85f182fdbbc..3e1aad5f1f9 100644
--- a/logstash-core/lib/logstash/java_pipeline.rb
+++ b/logstash-core/lib/logstash/java_pipeline.rb
@@ -455,6 +455,7 @@ def shutdown
     # TODO: should we also check against calling shutdown multiple times concurrently?
     stop_inputs
     wait_for_shutdown
+  ensure
     clear_pipeline_metrics
   end # def shutdown
 
diff --git a/logstash-core/spec/logstash/java_pipeline_spec.rb b/logstash-core/spec/logstash/java_pipeline_spec.rb
index 22915b85501..59700c86444 100644
--- a/logstash-core/spec/logstash/java_pipeline_spec.rb
+++ b/logstash-core/spec/logstash/java_pipeline_spec.rb
@@ -21,6 +21,7 @@
 require_relative "../support/mocks_classes"
 require_relative "../support/helpers"
 require 'support/pipeline/pipeline_helpers'
+require 'logstash/pipeline_action/reload'
 require "stud/try"
 require 'timeout'
 require "thread"
@@ -75,6 +76,23 @@ def push_once
   end
 end
 
+class DummyCrashingInput  < LogStash::Inputs::Base
+  config_name "dummycrashinginput"
+
+  config :on_register, :validate => :boolean
+
+  def register
+    raise(LogStash::ConfigurationError, 'crashing input on register') if @on_register
+  end
+
+  def run(queue)
+    while !stop?
+      queue << LogStash::Event.new
+      sleep(0.5)
+    end
+  end
+end
+
 class DummyCodec < LogStash::Codecs::Base
   config_name "dummycodec"
   milestone 2
@@ -1407,4 +1425,100 @@ def flush(options)
       end
     end
   end
+
+  context "#shutdown" do
+    subject(:pipeline) { mock_java_pipeline_from_string(config, pipeline_settings_obj, pipeline_metric) }
+    let(:pipeline_metric) { LogStash::Instrument::Metric.new(LogStash::Instrument::Collector.new) }
+    let(:dummyinput_class) { DummyManualInputGenerator }
+    let(:dummyinput_config) { {'enable_metric' => true} }
+    let(:dummyinput) { dummyinput_class.new(dummyinput_config) }
+    let(:pipeline_id) { "shutdown" }
+    let(:config) { build_pipeline_string_config(dummyinput_config) }
+
+    def build_pipeline_string_config(dummyinput_config)
+      <<-"EOS"
+      input { dummyinput { #{ dummyinput_config.to_s.tr('{},', '') } } }
+      filter { dummyfilter {} }
+      output { dummyoutput {} }
+      EOS
+    end
+
+    before(:each) do
+      allow(::LogStash::Outputs::DummyOutput).to receive(:new).with(any_args).and_return(::LogStash::Outputs::DummyOutput::new)
+      allow(DummyManualInputGenerator).to receive(:new).with(any_args).and_return(dummyinput)
+
+      allow(LogStash::Plugin).to receive(:lookup).with("input", "dummyinput").and_return(dummyinput_class)
+      allow(LogStash::Plugin).to receive(:lookup).with("codec", "plain").and_return(LogStash::Codecs::Plain)
+      allow(LogStash::Plugin).to receive(:lookup).with("filter", "dummyfilter").and_return(DummyFilter)
+      allow(LogStash::Plugin).to receive(:lookup).with("output", "dummyoutput").and_return(::LogStash::Outputs::DummyOutput)
+
+      allow(pipeline).to receive(:clear_pipeline_metrics).and_call_original
+      allow(pipeline).to receive(:stop_inputs).and_call_original
+      allow(pipeline).to receive(:wait_for_shutdown).and_return(double(new: nil))
+    end
+
+    context "of a running pipeline" do
+      let(:pipeline_settings) { { "pipeline.batch.size" => 1, "pipeline.workers" => 1, "pipeline.id" => pipeline_id, "metric.collect" => true } }
+
+      it "should clear the pipeline metrics" do
+        dummyinput.keep_running.make_true
+
+        expect { pipeline.start }.to_not raise_error
+
+        pipeline.shutdown
+
+        expect(pipeline).to have_received(:clear_pipeline_metrics)
+        expect(pipeline).to have_received(:stop_inputs)
+        expect(pipeline).to have_received(:wait_for_shutdown)
+      end
+    end
+
+    context "of a failed reloading pipeline" do
+      let(:agent) { double("agent") }
+      let(:dummyinput_class) { DummyCrashingInput }
+      let(:pipeline_settings) { {
+        "pipeline.id" => pipeline_id,
+        "pipeline.batch.size" => 1,
+        "pipeline.workers" => 1,
+        "metric.collect" => true,
+        "config.reload.automatic" => true
+      } }
+      let(:pipelines) do
+        registry = LogStash::PipelinesRegistry.new
+        registry.create_pipeline(pipeline_id.to_sym, pipeline) { true }
+        registry
+      end
+
+      # This test ensure that even for failed pipeline reloads, the shutdown method cleans
+      # the pipeline's metric store. If those metrics are not cleaned, it would pilling up
+      # data from previous reloads attempts.
+      it "should clear the pipeline metrics for every shutdown" do
+        expect { pipeline.start }.to_not raise_error
+
+        wait(10).for { pipeline.running? }
+
+        new_config_string = build_pipeline_string_config({'on_register' => 'true' })
+        reload_action = LogStash::PipelineAction::Reload.new(
+          mock_pipeline_config(pipeline_id, new_config_string, pipeline_settings),
+          pipeline_metric
+        )
+
+        # 1st failed attempt (finished_execution? = > false)
+        action_result = reload_action.execute(agent, pipelines)
+        expect(action_result.successful?).to be_falsy
+        expect(pipeline).to have_received(:clear_pipeline_metrics)
+        expect(pipeline).to have_received(:stop_inputs)
+        expect(pipeline).to have_received(:wait_for_shutdown)
+
+        # 2nd failed attempt (finished_execution? = > true, as the previous reload failed)
+        pipeline = pipelines.states.get(pipeline_id.to_sym).pipeline
+        allow(pipeline).to receive(:wait_for_shutdown).and_return(double(new: nil))
+        allow(pipeline).to receive(:clear_pipeline_metrics).and_call_original
+
+        action_result = reload_action.execute(agent, pipelines)
+        expect(action_result.successful?).to be_falsy
+        expect(pipeline).to have_received(:clear_pipeline_metrics)
+      end
+    end
+  end
 end