[AGNTLOG-182] feat: add site config option

README.md

@@ -100,7 +100,8 @@ As `fluent-plugin-datadog` is a buffered output plugin, you can set all of the b
 | **dd_hostname** | Used by Datadog to identify the host submitting the logs. | `hostname -f` |
 | **service** | Used by Datadog to correlate between logs, traces and metrics. | nil |
 | **port** | Proxy port when logs are not directly forwarded to Datadog and ssl is not used | 80 |
-| **host** | Proxy endpoint when logs are not directly forwarded to Datadog | http-intake.logs.datadoghq.com |
+| **site** | The Datadog [site](https://docs.datadoghq.com/getting_started/site/) to send logs to. Used to derive the default `host` when no explicit `host` is provided. Valid values: `datadoghq.com`, `datadoghq.eu`, `us3.datadoghq.com`, `us5.datadoghq.com`, `ap1.datadoghq.com`, `ddog-gov.com`. | datadoghq.com |
+| **host** | Proxy endpoint when logs are not directly forwarded to Datadog. When unset, the default is derived from `site` as `http-intake.logs.<site>`. An explicitly configured `host` always wins over `site`. | (derived from `site`) |
 | **http_proxy** | HTTP proxy, only takes effect if HTTP forwarding is enabled (`use_http`). Defaults to `HTTP_PROXY`/`http_proxy` env vars. | nil |
 | **delete_extracted_tag_attributes** | When true, removes `kubernetes` and `docker` attributes from log records after extracting them as tags. Useful to avoid duplicate data in Datadog logs UI. | false |
 

lib/fluent/plugin/out_datadog.rb

@@ -24,7 +24,9 @@ class RetryableError < StandardError;
   DD_MAX_BATCH_SIZE = 5000000
   DD_TRUNCATION_SUFFIX = "...TRUNCATED..."
 
-  DD_DEFAULT_HTTP_ENDPOINT = "http-intake.logs.datadoghq.com"
+  DD_DEFAULT_SITE = "datadoghq.com"
+  DD_DEFAULT_HTTP_HOST_PREFIX = "http-intake.logs."
+  DD_DEFAULT_HTTP_ENDPOINT = "#{DD_DEFAULT_HTTP_HOST_PREFIX}#{DD_DEFAULT_SITE}".freeze
   DD_DEFAULT_TCP_ENDPOINT = "intake.logs.datadoghq.com"
 
   helpers :compat_parameters
@@ -45,8 +47,16 @@ class RetryableError < StandardError;
   config_param :dd_hostname, :string, :default => nil
   config_param :delete_extracted_tag_attributes, :bool, :default => false
 
+  # Datadog site used to derive the default intake host. Valid values include:
+  # "datadoghq.com" (default), "datadoghq.eu", "us3.datadoghq.com",
+  # "us5.datadoghq.com", "ap1.datadoghq.com", "ddog-gov.com". Any value
+  # explicitly set for `host` takes precedence over the site-derived default.
+  config_param :site, :string, :default => DD_DEFAULT_SITE
+
   # Connection settings
-  config_param :host, :string, :default => DD_DEFAULT_HTTP_ENDPOINT
+  # `host` defaults to nil so we can tell whether the user explicitly set it.
+  # When nil, the host is derived from `site` during `configure`.
+  config_param :host, :string, :default => nil
   config_param :use_ssl, :bool, :default => true
   config_param :port, :integer, :default => 80
   config_param :ssl_port, :integer, :default => 443
@@ -77,12 +87,29 @@ def initialize
   def configure(conf)
     compat_parameters_convert(conf, :buffer)
     super
-    return if @dd_hostname
 
-    if not @use_http and @host == DD_DEFAULT_HTTP_ENDPOINT
-      @host = DD_DEFAULT_TCP_ENDPOINT
+    # Derive default host from `site` only for HTTP transport.
+    # TCP users with a non-default site must set `host` explicitly;
+    # TCP users with no host and the default site get the legacy TCP endpoint.
+    if @host.nil? || @host.empty?
+      if @use_http
+        @host = "#{DD_DEFAULT_HTTP_HOST_PREFIX}#{@site}"
+      elsif @site == DD_DEFAULT_SITE
+        @host = DD_DEFAULT_TCP_ENDPOINT
+      else
+        # TCP + non-default site: we cannot safely derive a TCP intake hostname
+        # from `site` (the HTTP-intake prefix is HTTP-only), and leaving @host
+        # nil would surface as a cryptic connect-time error. Fail fast at
+        # configure time with an actionable message.
+        raise Fluent::ConfigError,
+          "`host` is required when `use_http false` is combined with a non-default `site` " \
+          "(`site #{@site.inspect}`). Set `host` explicitly to your TCP intake (e.g. " \
+          "`intake.logs.#{@site}`)."
+      end
     end
 
+    return if @dd_hostname
+
     # Set dd_hostname if not already set (can be set when using fluentd as aggregator)
     @dd_hostname = %x[hostname -f 2> /dev/null].strip
     @dd_hostname = Socket.gethostname if @dd_hostname.empty?

test/plugin/test_out_datadog.rb

@@ -82,6 +82,103 @@ def create_valid_subject
         end
       end
     end
+
+    test "default site yields the default HTTP host" do
+      plugin = create_driver(%[
+        api_key foo
+      ]).instance
+      assert_equal "datadoghq.com", plugin.site
+      assert_equal "http-intake.logs.datadoghq.com", plugin.host
+    end
+
+    test "EU site overrides the default HTTP host" do
+      plugin = create_driver(%[
+        api_key foo
+        site datadoghq.eu
+      ]).instance
+      assert_equal "datadoghq.eu", plugin.site
+      assert_equal "http-intake.logs.datadoghq.eu", plugin.host
+    end
+
+    test "explicit host overrides site-derived default" do
+      plugin = create_driver(%[
+        api_key foo
+        site datadoghq.eu
+        host my-custom-intake.example.com
+      ]).instance
+      assert_equal "datadoghq.eu", plugin.site
+      assert_equal "my-custom-intake.example.com", plugin.host
+    end
+
+    test "TCP transport with non-default site does not derive an HTTP host" do
+      # When use_http is false and a non-default site is set without an explicit
+      # host, @host must remain nil/empty so the caller is forced to set it
+      # explicitly. The old code would silently set an HTTP intake hostname
+      # and hand it to the TCP client.
+      plugin = create_driver(%[
+        api_key foo
+        use_http false
+        site datadoghq.eu
+        host intake.logs.datadoghq.eu
+      ]).instance
+      # The explicit host is used as-is; the HTTP-intake prefix must NOT appear.
+      assert_equal "intake.logs.datadoghq.eu", plugin.host
+      assert_false plugin.host.start_with?("http-intake.logs.")
+    end
+
+    test "TCP transport with no host and default site falls back to TCP endpoint" do
+      # Legacy behaviour: TCP + no host + default site => DD_DEFAULT_TCP_ENDPOINT
+      plugin = create_driver(%[
+        api_key foo
+        use_http false
+      ]).instance
+      assert_equal Fluent::DatadogOutput::DD_DEFAULT_TCP_ENDPOINT, plugin.host
+    end
+
+    test "TCP transport with non-default site and no host raises ConfigError" do
+      # This is the genuinely new and risky case: `use_http false` + a non-default
+      # `site` + no explicit `host`. The HTTP-intake prefix cannot be used for TCP,
+      # and we want a clear error at configure time instead of a cryptic connect-time
+      # failure later.
+      assert_raise(Fluent::ConfigError) do
+        create_driver(%[
+          api_key foo
+          use_http false
+          site datadoghq.eu
+        ])
+      end
+    end
+
+    test "gov site is accepted" do
+      plugin = create_driver(%[
+        api_key foo
+        site ddog-gov.com
+      ]).instance
+      assert_equal "ddog-gov.com", plugin.site
+      assert_equal "http-intake.logs.ddog-gov.com", plugin.host
+    end
+
+    test "unknown site is accepted (no validation, matches agent behavior)" do
+      # The Datadog Agent does not validate `site` (see pkg/config/utils/endpoints.go:
+      # GetMainEndpoint uses `prefix + strings.TrimSpace(c.GetString("site"))` directly).
+      # Mirror that behaviour here — typos surface at DNS resolution time, consistent
+      # with the rest of the Datadog product surface.
+      plugin = create_driver(%[
+        api_key foo
+        site future-region.datadoghq.com
+      ]).instance
+      assert_equal "future-region.datadoghq.com", plugin.site
+      assert_equal "http-intake.logs.future-region.datadoghq.com", plugin.host
+    end
+
+    test "valid subdomain site is accepted" do
+      plugin = create_driver(%[
+        api_key foo
+        site us3.datadoghq.com
+      ]).instance
+      assert_equal "us3.datadoghq.com", plugin.site
+      assert_equal "http-intake.logs.us3.datadoghq.com", plugin.host
+    end
   end
 
   sub_test_case "enrich_record" do