statsd-instrument 3.9.4 → 3.9.5

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 8248b85fead25773388f7bec728c5a6347f7b7af61e37e064ee8ed93ba31397d
-  data.tar.gz: 110ab2e5db1b66ef7c76e11ed7b82c925f767df0b09a1ca442afe95c31e30a40
+  metadata.gz: 90a8111f77847ab11ea592d7b4a15c91a1d52b1f706b9296f29afa0c943658e8
+  data.tar.gz: 444ef926b623f13fa9cb453b4179969a3655071a20c39b0f429dee3f22272662
 SHA512:
-  metadata.gz: e85d2333cae7f6e843b25b3a1e64cf8bd1e68e47bcb530b6483dfa05e2672374791209c65f888cc6c9354f0e5bca997098f776c5db10f4dfd155005dd3f22bc9
-  data.tar.gz: 330d5d152a42dd76fae018e4af8c4daa64f9be50391468f64bd338ddd91ecd668a6775f590e45452b552c6fe28f0e82e125a3a91476d5269e52d19ad0adee092
+  metadata.gz: ead57ab89be89a51dabe4eb08c93fdd73cb5a439d5605f38305268b81e079a62c18555e1cf462ce1aa231b1928301dfe62376fa31f073c419408e478dd5cfef3
+  data.tar.gz: 372737145878f6b8eda3c1e2d498fd40dbb7f5aa38acf13e94b93b2ace670e2bf787162dcf54b1489fb9d057ea76511b07a83b03a3bbcc740edf1c64081b403f

data/CHANGELOG.md

@@ -6,6 +6,11 @@ section below.
 
 ## Unreleased changes
 
+## Version 3.9.5
+
+- [#387](https://github.com/Shopify/statsd-instrument/pull/387) - Fixing bug when using aggregation and sending sampled
+histogram metrics, they will not be scaled properly because of missing sampling rate in the final sent sample.
+
 ## Version 3.9.4
 
 - [#384](https://github.com/Shopify/statsd-instrument/pull/384) - Aggregation: fixing bug when sending metrics synchronously

data/lib/statsd/instrument/aggregator.rb

@@ -3,13 +3,14 @@
 module StatsD
   module Instrument
     class AggregationKey
-      attr_reader :name, :tags, :no_prefix, :type, :hash
+      attr_reader :name, :tags, :no_prefix, :type, :hash, :sample_rate
 
-      def initialize(name, tags, no_prefix, type)
+      def initialize(name, tags, no_prefix, type, sample_rate: 1.0)
         @name = name
         @tags = tags
         @no_prefix = no_prefix
         @type = type
+        @sample_rate = sample_rate
         @hash = [@name, @tags, @no_prefix, @type].hash
       end
 
@@ -56,7 +57,7 @@ module StatsD
                   key.name,
                   key.type.to_s,
                   agg_value,
-                  CONST_SAMPLE_RATE,
+                  key.sample_rate,
                   key.tags,
                 )
               when GAUGE
@@ -134,7 +135,7 @@ module StatsD
         end
       end
 
-      def aggregate_timing(name, value, tags: [], no_prefix: false, type: DISTRIBUTION)
+      def aggregate_timing(name, value, tags: [], no_prefix: false, type: DISTRIBUTION, sample_rate: CONST_SAMPLE_RATE)
         unless thread_healthcheck
           @sink << datagram_builder(no_prefix: no_prefix).timing_value_packed(
             name, type.to_s, [value], CONST_SAMPLE_RATE, tags
@@ -143,7 +144,7 @@ module StatsD
         end
 
         tags = tags_sorted(tags)
-        key = packet_key(name, tags, no_prefix, type)
+        key = packet_key(name, tags, no_prefix, type, sample_rate: sample_rate)
 
         @mutex.synchronize do
           values = @aggregation_state[key] ||= []
@@ -176,6 +177,9 @@ module StatsD
 
       EMPTY_ARRAY = [].freeze
 
+      # Flushes the aggregated metrics to the sink.
+      # Iterates over the aggregation state and sends each metric to the sink.
+      # If you change this function, you need to update the logic in the finalizer as well.
       def do_flush
         @aggregation_state.each do |key, value|
           case key.type
@@ -191,7 +195,7 @@ module StatsD
               key.name,
               key.type.to_s,
               value,
-              CONST_SAMPLE_RATE,
+              key.sample_rate,
               key.tags,
             )
           when GAUGE
@@ -219,8 +223,14 @@ module StatsD
         datagram_builder(no_prefix: false).normalize_tags(tags)
       end
 
-      def packet_key(name, tags = "".b, no_prefix = false, type = COUNT)
-        AggregationKey.new(DatagramBuilder.normalize_string(name), tags, no_prefix, type).freeze
+      def packet_key(name, tags = "".b, no_prefix = false, type = COUNT, sample_rate: CONST_SAMPLE_RATE)
+        AggregationKey.new(
+          DatagramBuilder.normalize_string(name),
+          tags,
+          no_prefix,
+          type,
+          sample_rate: sample_rate,
+        ).freeze
       end
 
       def datagram_builder(no_prefix:)

data/lib/statsd/instrument/version.rb

@@ -2,6 +2,6 @@
 
 module StatsD
   module Instrument
-    VERSION = "3.9.4"
+    VERSION = "3.9.5"
   end
 end

data/test/aggregator_test.rb

@@ -58,6 +58,23 @@ class AggregatorTest < Minitest::Test
     assert_equal([1.0, 100.0], datagram.value)
   end
 
+  def test_timing_sampling_scaling
+    @subject.aggregate_timing("timing.sampled", 60.0, sample_rate: 0.01)
+    @subject.aggregate_timing("timing.sampled", 80.0, sample_rate: 0.01)
+    @subject.aggregate_timing("timing.unsampled", 60.0, sample_rate: 1.0)
+
+    @subject.flush
+
+    assert_equal(2, @sink.datagrams.size)
+
+    sampled_datagram = @sink.datagrams.find { |d| d.name == "timing.sampled" }
+    assert_equal([60.0, 80.0], sampled_datagram.value)
+    assert_equal(0.01, sampled_datagram.sample_rate)
+
+    unsampled_datagram = @sink.datagrams.find { |d| d.name == "timing.unsampled" }
+    assert_equal(60.0, unsampled_datagram.value)
+  end
+
   def test_mixed_type_timings
     @subject.aggregate_timing("foo_ms", 1, tags: { foo: "bar" }, type: :ms)
     @subject.aggregate_timing("foo_ms", 100, tags: { foo: "bar" }, type: :ms)
@@ -304,6 +321,7 @@ class AggregatorTest < Minitest::Test
     @subject.increment("foo", 1, tags: { foo: "bar" })
     @subject.aggregate_timing("bar", 100, tags: { foo: "bar" })
     @subject.gauge("baz", 100, tags: { foo: "bar" })
+    @subject.aggregate_timing("sampled_timing", 100, tags: { foo: "bar" }, sample_rate: 0.01)
 
     # Manually trigger the finalizer
     finalizer = StatsD::Instrument::Aggregator.finalize(
@@ -316,7 +334,7 @@ class AggregatorTest < Minitest::Test
     finalizer.call
 
     # Verify that all pending metrics are sent
-    assert_equal(3, @sink.datagrams.size)
+    assert_equal(4, @sink.datagrams.size)
 
     counter_datagram = @sink.datagrams.find { |d| d.name == "foo" }
     assert_equal(1, counter_datagram.value)
@@ -329,5 +347,9 @@ class AggregatorTest < Minitest::Test
     gauge_datagram = @sink.datagrams.find { |d| d.name == "baz" }
     assert_equal(100, gauge_datagram.value)
     assert_equal(["foo:bar"], gauge_datagram.tags)
+
+    sampled_timing_datagram = @sink.datagrams.find { |d| d.name == "sampled_timing" }
+    assert_equal(100.0, sampled_timing_datagram.value)
+    assert_equal(0.01, sampled_timing_datagram.sample_rate)
   end
 end

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: statsd-instrument
 version: !ruby/object:Gem::Version
-  version: 3.9.4
+  version: 3.9.5
 platform: ruby
 authors:
 - Jesse Storimer
@@ -10,7 +10,7 @@ authors:
 autorequire:
 bindir: bin
 cert_chain: []
-date: 2024-10-29 00:00:00.000000000 Z
+date: 2024-10-30 00:00:00.000000000 Z
 dependencies: []
 description: A StatsD client for Ruby apps. Provides metaprogramming methods to inject
   StatsD instrumentation into your code.