logstash-output-scalyr 0.1.25.beta → 0.2.1.beta

data/spec/benchmarks/json_serialization.rb

@@ -0,0 +1,85 @@
+require 'benchmark'
+require 'json'
+require 'jrjackson'
+
+require_relative '../../lib/scalyr/common/util'
+require_relative './util'
+
+ITERATIONS = 500
+
+def json_serialize_data_native(data)
+  data.to_json
+end
+
+def json_serialize_data_jrjackson(data)
+  JrJackson::Json.dump(data)
+end
+
+DATASETS = {
+  :keys_50 => generate_data_array_for_spec([3, 3, 3, 2]),
+  :keys_200 => generate_data_array_for_spec([4, 4, 3, 4]),
+  :keys_200_flat => generate_data_array_for_spec([200]),
+  :keys_512 => generate_data_array_for_spec([8, 4, 4, 4]),
+  :keys_960 => generate_data_array_for_spec([12, 5, 4, 4]),
+  :keys_2700 => generate_data_array_for_spec([14, 8, 6, 4])
+}
+
+def run_benchmark_and_print_results(data, run_benchmark_func)
+  puts ""
+  puts "Using %s total keys in a hash" % [Scalyr::Common::Util.flatten(data[0]).count]
+  puts ""
+
+  result = []
+  ITERATIONS.times do |i|
+    result << Benchmark.measure { run_benchmark_func.(data[i]) }
+  end
+
+  sum = result.inject(nil) { |sum, t| sum.nil? ? sum = t : sum += t }
+  avg = sum / result.size
+
+  Benchmark.bm(7, "sum:", "avg:") do |b|
+    [sum, avg]
+  end
+  puts ""
+end
+
+puts "Using %s iterations" % [ITERATIONS]
+puts ""
+
+puts "native"
+puts "==============================="
+
+# Around ~50 keys in a hash
+data = DATASETS[:keys_50]
+run_benchmark_and_print_results(data, method(:json_serialize_data_native))
+
+# Around ~200 keys in a hash
+data = DATASETS[:keys_200]
+run_benchmark_and_print_results(data, method(:json_serialize_data_native))
+
+# Around ~200 keys in a hash (single level)
+data = DATASETS[:keys_200_flat]
+run_benchmark_and_print_results(data, method(:json_serialize_data_native))
+
+# Around ~2700 keys in a hash
+data = DATASETS[:keys_2700]
+run_benchmark_and_print_results(data, method(:json_serialize_data_native))
+
+puts "jrjackson"
+puts "==============================="
+
+# Around ~50 keys in a hash
+data = DATASETS[:keys_50]
+run_benchmark_and_print_results(data, method(:json_serialize_data_jrjackson))
+
+# Around ~200 keys in a hash
+data = DATASETS[:keys_200]
+run_benchmark_and_print_results(data, method(:json_serialize_data_jrjackson))
+
+# Around ~200 keys in a hash (single level)
+data = DATASETS[:keys_200_flat]
+run_benchmark_and_print_results(data, method(:json_serialize_data_jrjackson))
+
+# Around ~2700 keys in a hash
+data = DATASETS[:keys_2700]
+run_benchmark_and_print_results(data, method(:json_serialize_data_jrjackson))

data/spec/benchmarks/set_session_level_serverhost_on_events.rb

@@ -0,0 +1,107 @@
+require 'benchmark'
+require 'quantile'
+
+require_relative '../../lib/scalyr/constants'
+require_relative '../../lib/scalyr/common/util'
+require_relative './util'
+
+# Micro benchmark which measures how long "set_session_level_serverhost_on_events" takes
+
+ITERATIONS = 100
+
+def run_benchmark_and_print_results(data, run_benchmark_func)
+  puts ""
+  puts "Using %s total events in a batch" % [data[0].size]
+  puts ""
+
+  result = []
+  ITERATIONS.times do |i|
+    result << Benchmark.measure { run_benchmark_func.(data[i]) }
+  end
+
+  sum = result.inject(nil) { |sum, t| sum.nil? ? sum = t : sum += t }
+  avg = sum / result.size
+
+  Benchmark.bm(7, "sum:", "avg:") do |b|
+    [sum, avg]
+  end
+  puts ""
+end
+
+# Generate random event with only single event having special server host attribute set which
+# represents a worst case scenario since we need to backfill rest of the events.
+def generate_events(count)
+  result = []
+
+  ITERATIONS.times do |iteration|
+    events = []
+
+    count.times do |index|
+      event = generate_hash([2])
+      event[:attrs] = Hash.new
+      event[:log] = 1
+
+      if index == count - 1
+        event[:attrs][EVENT_LEVEL_SERVER_HOST_ATTRIBUTE_NAME] = format("test-host-%s", index)
+      end
+
+      events << event
+    end
+
+    raise "Assertion failed" unless events.size == count
+
+    result << events
+  end
+
+  raise "Assertion failed" unless result.size == ITERATIONS
+  result
+end
+
+def run_func(events)
+  # NOTE: This function manipulates events in place
+  events.each_with_index do |event, index|
+    if index < events.size - 1
+      # Last event will have _origServerHost set, but others won't
+      raise "Assertion failed" unless event[:attrs][EVENT_LEVEL_SERVER_HOST_ATTRIBUTE_NAME].nil?
+    end
+  end
+
+  Scalyr::Common::Util.set_session_level_serverhost_on_events("session-server-host-dummy", events, {}, true)
+
+  events.each do |event|
+    raise "Assertion failed" unless event[:attrs][EVENT_LEVEL_SERVER_HOST_ATTRIBUTE_NAME].nil? == false
+  end
+end
+
+
+puts "Using %s iterations" % [ITERATIONS]
+puts ""
+
+@value = Quantile::Estimator.new
+
+puts "Util.set_session_level_serverhost_on_events()"
+puts "==============================="
+
+# 100 events in a batch
+data = generate_events(100)
+run_benchmark_and_print_results(data, method(:run_func))
+
+# 500 events in a batch
+data = generate_events(500)
+run_benchmark_and_print_results(data, method(:run_func))
+
+# 1000 events in a batch
+data = generate_events(1000)
+run_benchmark_and_print_results(data, method(:run_func))
+
+# 2000 events in a batch
+data = generate_events(2000)
+run_benchmark_and_print_results(data, method(:run_func))
+
+# 3000 events in a batch
+data = generate_events(3000)
+run_benchmark_and_print_results(data, method(:run_func))
+
+# 5000 events in a batch
+data = generate_events(5000)
+run_benchmark_and_print_results(data, method(:run_func))

data/spec/benchmarks/util.rb

@@ -0,0 +1,24 @@
+def rand_str(len)
+  return (0...len).map { (65 + rand(26)).chr }.join
+end
+
+def generate_hash(widths)
+  result = {}
+  if widths.empty?
+    return rand_str(20)
+  else
+    widths[0].times do
+      result[rand_str(9)] = generate_hash(widths[1..widths.length])
+    end
+    return result
+  end
+end
+
+def generate_data_array_for_spec(spec)
+  data = []
+  ITERATIONS.times do
+    data << generate_hash(spec)
+  end
+
+  data
+end

data/spec/logstash/outputs/scalyr_integration_spec.rb

@@ -5,6 +5,11 @@ require "logstash/codecs/plain"
 require "logstash/event"
 require "json"
 require 'webmock/rspec'
+
+# Require the specific version of `json` used in logstash
+gem 'json', '1.8.6'
+require 'json'
+
 WebMock.allow_net_connect!
 
 RSpec.configure do |rspec|
@@ -31,7 +36,10 @@ describe LogStash::Outputs::Scalyr do
   describe "#ssl_tests" do
       context "with default SSL configuration" do
         it "throws a ServerError due to fake api key" do
-              plugin = LogStash::Outputs::Scalyr.new({'api_write_token' => '1234'})
+              plugin = LogStash::Outputs::Scalyr.new({
+                'api_write_token' => '1234',
+                'perform_connectivity_check' => false,
+              })
               plugin.register
               plugin.instance_variable_set(:@running, false)
               allow(plugin.instance_variable_get(:@logger)).to receive(:error)
@@ -42,7 +50,7 @@ describe LogStash::Outputs::Scalyr do
                   :batch_num=>1,
                   :code=>401,
                   :message=>"error/client/badParam",
-                  :payload_size=>781,
+                  :payload_size=>737,
                   :record_count=>3,
                   :total_batches=>1,
                   :url=>"https://agent.scalyr.com/addEvents",
@@ -55,7 +63,12 @@ describe LogStash::Outputs::Scalyr do
 
       context "when pointing at a location without any valid certs and not using builtin" do
         it "throws an SSLError" do
-              plugin = LogStash::Outputs::Scalyr.new({'api_write_token' => '1234', 'ssl_ca_bundle_path' => '/fakepath/nocerts', 'append_builtin_cert' => false})
+              plugin = LogStash::Outputs::Scalyr.new({
+                'api_write_token' => '1234',
+                'perform_connectivity_check' => false,
+                'ssl_ca_bundle_path' => '/fakepath/nocerts',
+                'append_builtin_cert' => false,
+              })
               plugin.register
               plugin.instance_variable_set(:@running, false)
               allow(plugin.instance_variable_get(:@logger)).to receive(:error)
@@ -65,7 +78,7 @@ describe LogStash::Outputs::Scalyr do
                   :error_class=>"Manticore::UnknownException",
                   :batch_num=>1,
                   :message=>"java.lang.RuntimeException: Unexpected error: java.security.InvalidAlgorithmParameterException: the trustAnchors parameter must be non-empty",
-                  :payload_size=>781,
+                  :payload_size=>737,
                   :record_count=>3,
                   :total_batches=>1,
                   :url=>"https://agent.scalyr.com/addEvents",
@@ -81,7 +94,11 @@ describe LogStash::Outputs::Scalyr do
           `sudo mv #{OpenSSL::X509::DEFAULT_CERT_DIR} /tmp/system_certs`
 
           begin
-              plugin = LogStash::Outputs::Scalyr.new({'api_write_token' => '1234', 'append_builtin_cert' => false})
+              plugin = LogStash::Outputs::Scalyr.new({
+                'api_write_token' => '1234',
+                'perform_connectivity_check' => false,
+                'append_builtin_cert' => false,
+              })
               plugin.register
               plugin.instance_variable_set(:@running, false)
               allow(plugin.instance_variable_get(:@logger)).to receive(:error)
@@ -91,7 +108,7 @@ describe LogStash::Outputs::Scalyr do
                   :error_class=>"Manticore::UnknownException",
                   :batch_num=>1,
                   :message=>"java.lang.RuntimeException: Unexpected error: java.security.InvalidAlgorithmParameterException: the trustAnchors parameter must be non-empty",
-                  :payload_size=>781,
+                  :payload_size=>737,
                   :record_count=>3,
                   :total_batches=>1,
                   :url=>"https://agent.scalyr.com/addEvents",
@@ -120,7 +137,11 @@ describe LogStash::Outputs::Scalyr do
           `echo "#{etc_hosts_entry}" | sudo tee -a /etc/hosts`
 
           begin
-              plugin = LogStash::Outputs::Scalyr.new({'api_write_token' => '1234', 'scalyr_server' => 'https://invalid.mitm.should.fail.test.agent.scalyr.com:443'})
+              plugin = LogStash::Outputs::Scalyr.new({
+                'api_write_token' => '1234',
+                'perform_connectivity_check' => false,
+                'scalyr_server' => 'https://invalid.mitm.should.fail.test.agent.scalyr.com:443',
+              })
               plugin.register
               plugin.instance_variable_set(:@running, false)
               allow(plugin.instance_variable_get(:@logger)).to receive(:error)
@@ -130,7 +151,7 @@ describe LogStash::Outputs::Scalyr do
                   :error_class=>"Manticore::UnknownException",
                   :batch_num=>1,
                   :message=>"Host name 'invalid.mitm.should.fail.test.agent.scalyr.com' does not match the certificate subject provided by the peer (CN=*.scalyr.com)",
-                  :payload_size=>781,
+                  :payload_size=>737,
                   :record_count=>3,
                   :total_batches=>1,
                   :url=>"https://invalid.mitm.should.fail.test.agent.scalyr.com/addEvents",
@@ -147,7 +168,13 @@ describe LogStash::Outputs::Scalyr do
 
       context "when an error occurs with retries at 5" do
         it "exits after 5 retries and emits a log" do
-              plugin = LogStash::Outputs::Scalyr.new({'retry_initial_interval' => 0.1, 'api_write_token' => '1234', 'ssl_ca_bundle_path' => '/fakepath/nocerts', 'append_builtin_cert' => false})
+              plugin = LogStash::Outputs::Scalyr.new({
+                'api_write_token' => '1234',
+                'perform_connectivity_check' => false,
+                'retry_initial_interval' => 0.1,
+                'ssl_ca_bundle_path' => '/fakepath/nocerts',
+                'append_builtin_cert' => false
+              })
               plugin.register
               allow(plugin.instance_variable_get(:@logger)).to receive(:error)
               plugin.multi_receive(sample_events)
@@ -163,7 +190,12 @@ describe LogStash::Outputs::Scalyr do
         stub_request(:post, "https://agent.scalyr.com/addEvents").
           to_return(status: 503, body: "stubbed response", headers: {})
 
-        plugin = LogStash::Outputs::Scalyr.new({'api_write_token' => '1234', 'ssl_ca_bundle_path' => '/fakepath/nocerts', 'append_builtin_cert' => false})
+        plugin = LogStash::Outputs::Scalyr.new({
+          'api_write_token' => '1234',
+          'perform_connectivity_check' => false,
+          'ssl_ca_bundle_path' => '/fakepath/nocerts',
+          'append_builtin_cert' => false
+        })
         plugin.register
         plugin.instance_variable_set(:@running, false)
 
@@ -175,7 +207,7 @@ describe LogStash::Outputs::Scalyr do
             :batch_num=>1,
             :code=>503,
             :message=>"Invalid JSON response from server",
-            :payload_size=>781,
+            :payload_size=>737,
             :record_count=>3,
             :total_batches=>1,
             :url=>"https://agent.scalyr.com/addEvents",
@@ -191,7 +223,12 @@ describe LogStash::Outputs::Scalyr do
         stub_request(:post, "https://agent.scalyr.com/addEvents").
           to_return(status: 500, body: "stubbed response", headers: {})
 
-        plugin = LogStash::Outputs::Scalyr.new({'api_write_token' => '1234', 'ssl_ca_bundle_path' => '/fakepath/nocerts', 'append_builtin_cert' => false})
+        plugin = LogStash::Outputs::Scalyr.new({
+          'api_write_token' => '1234',
+          'perform_connectivity_check' => false,
+          'ssl_ca_bundle_path' => '/fakepath/nocerts',
+          'append_builtin_cert' => false
+        })
         plugin.register
         plugin.instance_variable_set(:@running, false)
 
@@ -203,7 +240,7 @@ describe LogStash::Outputs::Scalyr do
             :batch_num=>1,
             :code=>500,
             :message=>"Invalid JSON response from server",
-            :payload_size=>781,
+            :payload_size=>737,
             :record_count=>3,
             :total_batches=>1,
             :url=>"https://agent.scalyr.com/addEvents",
@@ -219,7 +256,12 @@ describe LogStash::Outputs::Scalyr do
         stub_request(:post, "https://agent.scalyr.com/addEvents").
           to_return(status: 500, body: "0123456789" * 52, headers: {})
 
-        plugin = LogStash::Outputs::Scalyr.new({'api_write_token' => '1234', 'ssl_ca_bundle_path' => '/fakepath/nocerts', 'append_builtin_cert' => false})
+        plugin = LogStash::Outputs::Scalyr.new({
+            'api_write_token' => '1234',
+            'perform_connectivity_check' => false,
+            'ssl_ca_bundle_path' => '/fakepath/nocerts',
+            'append_builtin_cert' => false
+        })
         plugin.register
         plugin.instance_variable_set(:@running, false)
 
@@ -231,7 +273,7 @@ describe LogStash::Outputs::Scalyr do
             :batch_num=>1,
             :code=>500,
             :message=>"Invalid JSON response from server",
-            :payload_size=>781,
+            :payload_size=>737,
             :record_count=>3,
             :total_batches=>1,
             :url=>"https://agent.scalyr.com/addEvents",
@@ -248,7 +290,12 @@ describe LogStash::Outputs::Scalyr do
         stub_request(:post, "https://agent.scalyr.com/addEvents").
           to_return(status: 500, body: "stubbed response", headers: {})
 
-        plugin = LogStash::Outputs::Scalyr.new({'api_write_token' => '1234', 'ssl_ca_bundle_path' => '/fakepath/nocerts', 'append_builtin_cert' => false})
+        plugin = LogStash::Outputs::Scalyr.new({
+            'api_write_token' => '1234',
+            'perform_connectivity_check' => false,
+            'ssl_ca_bundle_path' => '/fakepath/nocerts',
+            'append_builtin_cert' => false
+        })
         plugin.register
         plugin.instance_variable_set(:@running, false)
         plugin.instance_variable_set('@dlq_writer', dlq_writer)