From 3ba57a04fcacae0a29cc45c4c128c0c940264391 Mon Sep 17 00:00:00 2001
From: Adrian Gruntkowski <adrian.gruntkowski@gmail.com>
Date: Mon, 13 Nov 2023 14:57:51 +0100
Subject: [PATCH] Make OpenTelemetry sampler ratio configurable via env (#3514)

---
 config/runtime.exs                      |  7 ++++++-
 lib/plausible/open_telemetry/sampler.ex | 23 ++++++++++++-----------
 2 files changed, 18 insertions(+), 12 deletions(-)

diff --git a/config/runtime.exs b/config/runtime.exs
index 7e92329d4..424496ea9 100644
--- a/config/runtime.exs
+++ b/config/runtime.exs
@@ -178,6 +178,11 @@ google_cid = get_var_from_path_or_env(config_dir, "GOOGLE_CLIENT_ID")
 google_secret = get_var_from_path_or_env(config_dir, "GOOGLE_CLIENT_SECRET")
 postmark_api_key = get_var_from_path_or_env(config_dir, "POSTMARK_API_KEY")
 
+{otel_sampler_ratio, ""} =
+  config_dir
+  |> get_var_from_path_or_env("OTEL_SAMPLER_RATIO", "0.5")
+  |> Float.parse()
+
 cron_enabled =
   config_dir
   |> get_var_from_path_or_env("CRON_ENABLED", "false")
@@ -612,7 +617,7 @@ end
 if honeycomb_api_key && honeycomb_dataset do
   config :opentelemetry,
     resource: Plausible.OpenTelemetry.resource_attributes(runtime_metadata),
-    sampler: {Plausible.OpenTelemetry.Sampler, nil},
+    sampler: {Plausible.OpenTelemetry.Sampler, %{ratio: otel_sampler_ratio}},
     span_processor: :batch,
     traces_exporter: :otlp
 
diff --git a/lib/plausible/open_telemetry/sampler.ex b/lib/plausible/open_telemetry/sampler.ex
index 53571159e..e6a459904 100644
--- a/lib/plausible/open_telemetry/sampler.ex
+++ b/lib/plausible/open_telemetry/sampler.ex
@@ -10,13 +10,10 @@ defmodule Plausible.OpenTelemetry.Sampler do
 
   import Bitwise, only: [&&&: 2]
 
-  # effective sampling ratio for non-ignored traces
-  @ratio 0.5
+  # mask for extracting first 64 bits of trace ID
   # 2^63 - 1
   @max_value 9_223_372_036_854_775_807
 
-  @id_upper_bound @ratio * @max_value
-
   @behaviour :otel_sampler
   require OpenTelemetry.Tracer, as: Tracer
 
@@ -24,13 +21,17 @@ defmodule Plausible.OpenTelemetry.Sampler do
   @tables_to_ignore ["oban_jobs"]
 
   @impl true
-  def setup(_sampler_opts), do: []
+  def setup(%{ratio: ratio}) when is_number(ratio) do
+    %{ratio: ratio, id_upper_bound: ratio * @max_value}
+  end
 
   @impl true
-  def description(_sampler_config), do: inspect(__MODULE__)
+  def description(%{ratio: ratio}) do
+    "#{inspect(__MODULE__)}{ratio=#{ratio}}"
+  end
 
   @impl true
-  def should_sample(context, trace_id, _links, _name, _kind, attributes, _config) do
+  def should_sample(context, trace_id, _links, _name, _kind, attributes, config) do
     tracestate = context |> Tracer.current_span_ctx() |> OpenTelemetry.Span.tracestate()
 
     case attributes do
@@ -41,19 +42,19 @@ defmodule Plausible.OpenTelemetry.Sampler do
         {:drop, [], tracestate}
 
       _any ->
-        {decide(trace_id), [], tracestate}
+        {decide(trace_id, config.id_upper_bound), [], tracestate}
     end
   end
 
-  defp decide(trace_id) when is_integer(trace_id) and trace_id > 0 do
+  defp decide(trace_id, id_upper_bound) when is_integer(trace_id) and trace_id > 0 do
     lower_64_bits = trace_id &&& @max_value
 
-    if abs(lower_64_bits) < @id_upper_bound do
+    if abs(lower_64_bits) < id_upper_bound do
       :record_and_sample
     else
       :drop
     end
   end
 
-  defp decide(_), do: :drop
+  defp decide(_, _), do: :drop
 end