feat(otelcol.receiver.datadog): Expose intake proxy and trace_id_cache_size settings

docs/sources/reference/components/otelcol/otelcol.receiver.datadog.md

@@ -49,6 +49,7 @@ You can use the following arguments with `otelcol.receiver.datadog`:
 | `keep_alives_enabled`    | `boolean`                  | Whether or not HTTP keep-alives are enabled                                  | `true`                                                     | no       |
 | `max_request_body_size`  | `string`                   | Maximum request body size the server will allow.                             | `"20MiB"`                                                  | no       |
 | `read_timeout`           | `duration`                 | Read timeout for requests of the HTTP server.                                | `"60s"`                                                    | no       |
+| `trace_id_cache_size`    | `int`                      | Cache size for mapping 64-bit to 128-bit trace IDs.                          | `0`                                                        | no       |
 
 
 
@@ -59,13 +60,16 @@ To expose the HTTP server to other machines on your network, configure `endpoint
 
 You can use the following blocks with `otelcol.receiver.datadog`:
 
-| Block                            | Description                                                                | Required |
-| -------------------------------- | -------------------------------------------------------------------------- | -------- |
-| [`output`][output]               | Configures where to send received telemetry data.                          | yes      |
-| [`cors`][cors]                   | Configures CORS for the HTTP server.                                       | no       |
-| [`debug_metrics`][debug_metrics] | Configures the metrics that this component generates to monitor its state. | no       |
-| [`tls`][tls]                     | Configures TLS for the HTTP server.                                        | no       |
-| `tls` > [`tpm`][tpm]             | Configures TPM settings for the TLS `key_file`.                            | no       |
+| Block                                       | Description                                                                | Required |
+| ------------------------------------------- | -------------------------------------------------------------------------- | -------- |
+| [`output`][output]                          | Configures where to send received telemetry data.                          | yes      |
+| [`cors`][cors]                              | Configures CORS for the HTTP server.                                       | no       |
+| [`debug_metrics`][debug_metrics]            | Configures the metrics that this component generates to monitor its state. | no       |
+| [`intake`][intake]                          | Configures the `/intake` endpoint behavior.                                | no       |
+| `intake` > [`proxy`][proxy]                 | Configures the proxy for the `/intake` endpoint.                           | no       |
+| `intake` > `proxy` > [`api`][api]           | Configures the Datadog API connection for the intake proxy.                | conditional |
+| [`tls`][tls]                                | Configures TLS for the HTTP server.                                        | no       |
+| `tls` > [`tpm`][tpm]                        | Configures TPM settings for the TLS `key_file`.                            | no       |
 
 The > symbol indicates deeper levels of nesting.
 For example, `tls` > `tpm` refers to a `tpm` block defined inside a `tls` block.
@@ -74,6 +78,9 @@ For example, `tls` > `tpm` refers to a `tpm` block defined inside a `tls` block.
 [tpm]: #tpm
 [cors]: #cors
 [debug_metrics]: #debug_metrics
+[intake]: #intake
+[proxy]: #proxy
+[api]: #api
 [output]: #output
 
 ### `output`
@@ -104,6 +111,41 @@ The following headers are always implicitly allowed:
 
 If `allowed_headers` includes `"*"`, all headers are permitted.
 
+### `intake`
+
+The `intake` block configures how the `/intake` endpoint behaves.
+The Datadog Agent uses this endpoint to submit host tags and other metadata.
+
+The following arguments are supported:
+
+| Name       | Type     | Description                                             | Default | Required |
+|------------|----------|---------------------------------------------------------|---------|----------|
+| `behavior` | `string` | How the `/intake` endpoint behaves: `"disable"` or `"proxy"`. | | yes |
+
+Set `behavior` to `"proxy"` to forward `/intake` requests to Datadog's API.
+Proxying requires a nested `proxy { api { ... } }` block with a Datadog API key.
+When set to `"disable"`, the endpoint returns an error for any incoming request.
+
+### `proxy`
+
+The `proxy` block configures how the `/intake` proxy operates.
+It's only used when `behavior` is set to `"proxy"`.
+If `behavior` isn't `"proxy"`, this block is ignored.
+
+This block has no arguments and is configured with the nested [`api`][api] block.
+
+### `api`
+
+The `api` block configures the Datadog API connection used by the intake proxy.
+
+The following arguments are supported:
+
+| Name                  | Type       | Description                                      | Default            | Required |
+|-----------------------|------------|--------------------------------------------------|--------------------|----------|
+| `key`                 | `secret`   | Datadog API key.                                 |                    | yes      |
+| `site`                | `string`   | Datadog site to send data to.                    | `"datadoghq.com"`  | no       |
+| `fail_on_invalid_key` | `bool`     | Exit on startup if the API key is invalid.       | `false`            | no       |
+
 ### `debug_metrics`
 
 {{< docs/shared lookup="reference/components/otelcol-debug-metrics-block.md" source="alloy" version="<ALLOY_VERSION>" >}}
@@ -159,6 +201,29 @@ otelcol.exporter.otlphttp "default" {
 }
 ```
 
+## Proxy `/intake` requests to Datadog
+
+You can configure the receiver to forward `/intake` requests (host tags and metadata) to Datadog's API while processing metrics and traces locally:
+
+```alloy
+otelcol.receiver.datadog "default" {
+  intake {
+    behavior = "proxy"
+    proxy {
+      api {
+        key  = sys.env("DD_API_KEY")
+        site = "datadoghq.eu"
+      }
+    }
+  }
+
+  output {
+    metrics = [otelcol.processor.batch.default.input]
+    traces  = [otelcol.processor.batch.default.input]
+  }
+}
+```
+
 ## Enable authentication
 
 You can create a `otelcol.receiver.datadog` component that requires authentication for requests.

internal/component/otelcol/receiver/datadog/datadog.go

@@ -2,15 +2,19 @@
 package datadog
 
 import (
+	"fmt"
 	"time"
 
 	"github.com/grafana/alloy/internal/component"
 	"github.com/grafana/alloy/internal/component/otelcol"
 	otelcolCfg "github.com/grafana/alloy/internal/component/otelcol/config"
 	"github.com/grafana/alloy/internal/component/otelcol/receiver"
 	"github.com/grafana/alloy/internal/featuregate"
+	"github.com/grafana/alloy/syntax/alloytypes"
+	datadogconfig "github.com/open-telemetry/opentelemetry-collector-contrib/pkg/datadog/config"
 	"github.com/open-telemetry/opentelemetry-collector-contrib/receiver/datadogreceiver"
 	otelcomponent "go.opentelemetry.io/collector/component"
+	"go.opentelemetry.io/collector/config/configopaque"
 	"go.opentelemetry.io/collector/pipeline"
 )
 
@@ -31,7 +35,10 @@ func init() {
 type Arguments struct {
 	HTTPServer otelcol.HTTPServerArguments `alloy:",squash"`
 
-	ReadTimeout time.Duration `alloy:"read_timeout,attr,optional"`
+	ReadTimeout      time.Duration `alloy:"read_timeout,attr,optional"`
+	TraceIDCacheSize int           `alloy:"trace_id_cache_size,attr,optional"`
+
+	Intake *IntakeArguments `alloy:"intake,block,optional"`
 
 	// DebugMetrics configures component internal metrics. Optional.
 	DebugMetrics otelcolCfg.DebugMetricsArguments `alloy:"debug_metrics,block,optional"`
@@ -40,8 +47,54 @@ type Arguments struct {
 	Output *otelcol.ConsumerArguments `alloy:"output,block"`
 }
 
+// IntakeArguments controls the /intake endpoint behavior.
+type IntakeArguments struct {
+	// Behavior is required; allowed values are "disable" or "proxy".
+	Behavior string          `alloy:"behavior,attr"`
+	Proxy    *ProxyArguments `alloy:"proxy,block,optional"`
+}
+
+// ProxyArguments controls how the /intake proxy operates.
+type ProxyArguments struct {
+	API APIArguments `alloy:"api,block"`
+}
+
+// APIArguments configures the Datadog API connection for the intake proxy.
+type APIArguments struct {
+	Key              alloytypes.Secret `alloy:"key,attr"`
+	Site             string            `alloy:"site,attr,optional"`
+	FailOnInvalidKey bool              `alloy:"fail_on_invalid_key,attr,optional"`
+}
+
 var _ receiver.Arguments = Arguments{}
 
+// Validate implements syntax.Validator.
+func (args *Arguments) Validate() error {
+	if args.Intake != nil {
+		if err := args.Intake.Validate(); err != nil {
+			return err
+		}
+	}
+
+	// Validate the converted upstream config. The upstream Validate() is not
+	// called automatically by the Alloy receiver framework, so we call it
+	// explicitly here. This also avoids duplicating upstream validation logic
+	// (e.g. allowed intake behavior values) which may evolve over time.
+	cfg, err := args.Convert()
+	if err != nil {
+		return err
+	}
+	return cfg.(*datadogreceiver.Config).Validate()
+}
+
+// Validate checks IntakeArguments constraints documented in the component reference.
+func (args *IntakeArguments) Validate() error {
+	if args.Behavior == "proxy" && args.Proxy == nil {
+		return fmt.Errorf("a proxy block with an api block is required when intake behavior is %q", args.Behavior)
+	}
+	return nil
+}
+
 // SetToDefault implements syntax.Defaulter.
 func (args *Arguments) SetToDefault() {
 	*args = Arguments{
@@ -61,10 +114,38 @@ func (args Arguments) Convert() (otelcomponent.Config, error) {
 		return nil, err
 	}
 
-	return &datadogreceiver.Config{
-		ServerConfig: *convertedHttpServer,
-		ReadTimeout:  args.ReadTimeout,
-	}, nil
+	cfg := &datadogreceiver.Config{
+		ServerConfig:     *convertedHttpServer,
+		ReadTimeout:      args.ReadTimeout,
+		TraceIDCacheSize: args.TraceIDCacheSize,
+	}
+
+	if args.Intake != nil {
+		cfg.Intake = args.Intake.Convert()
+	}
+
+	return cfg, nil
+}
+
+func (args *IntakeArguments) Convert() datadogreceiver.IntakeConfig {
+	ic := datadogreceiver.IntakeConfig{
+		Behavior: args.Behavior,
+	}
+	if args.Behavior == "proxy" && args.Proxy != nil {
+		apiSite := args.Proxy.API.Site
+		if apiSite == "" {
+			apiSite = datadogconfig.DefaultSite
+		}
+
+		ic.Proxy = datadogreceiver.ProxyConfig{
+			API: datadogconfig.APIConfig{
+				Key:              configopaque.String(args.Proxy.API.Key),
+				Site:             apiSite,
+				FailOnInvalidKey: args.Proxy.API.FailOnInvalidKey,
+			},
+		}
+	}
+	return ic
 }
 
 // Extensions implements receiver.Arguments.