Add config schemas for comp/otelcol components

comp/otelcol/.schemagen.yaml

@@ -0,0 +1,48 @@
+# Settings for upstream's schemagen tool.
+# https://github.com/open-telemetry/opentelemetry-collector-contrib/tree/main/cmd/schemagen
+#
+# To regenerate a component's config.schema.yaml:
+#
+#   1. Build the tool from a local checkout of opentelemetry-collector-contrib:
+#        (cd /path/to/opentelemetry-collector-contrib/cmd/schemagen && go build -o /tmp/schemagen .)
+#   2. Run it against the component's directory (the one containing config.go and metadata.yaml):
+#        /tmp/schemagen comp/otelcol/<component>/...
+#
+# The tool walks up from CWD to find this file, so it must remain at or above
+# every component path it generates schemas for. See ./README.md and individual
+# component metadata.yaml files for context.
+#
+# Known limitation: schemagen treats every Go-embedded field as squashed,
+# regardless of the explicit mapstructure tag value. Components with embedded
+# pointer fields (e.g. ddflareextension's *confighttp.ServerConfig) or embedded
+# fields with a non-squash mapstructure key (e.g. serializerexporter's
+# configtls.ClientConfig with mapstructure:"tls") need hand-tuning after
+# regeneration.
+
+namespace: github.com/DataDog/datadog-agent
+mappings:
+  time:
+    Time:
+      schemaType: string
+      format: date-time
+      skipAnnotation: true
+    Duration:
+      schemaType: string
+      format: duration
+      skipAnnotation: true
+  go.opentelemetry.io/collector/component:
+    ID:
+      schemaType: string
+  github.com/DataDog/datadog-agent/comp/core/tagger/types:
+    # TagCardinality is a typed int (iota: 0=low, 1=orchestrator, 2=high,
+    # 3=none, 4=checks-config). YAML accepts the integer form; there is no
+    # UnmarshalText hook today, so the schema mirrors that.
+    TagCardinality:
+      schemaType: integer
+allowedRefs:
+  - go.opentelemetry.io/collector
+  - github.com/open-telemetry/opentelemetry-collector-contrib
+  - github.com/DataDog/datadog-agent
+componentOverrides:
+  exporter/serializer:
+    configName: ExporterConfig

comp/otelcol/ddflareextension/impl/config.schema.yaml

@@ -0,0 +1,5 @@
+description: Config has the configuration for the extension enabling the health check extension, used to report the health status of the service.
+type: object
+allOf:
+  - x-pointer: true
+    $ref: go.opentelemetry.io/collector/config/confighttp.server_config

comp/otelcol/ddflareextension/impl/metadata.yaml

@@ -0,0 +1,9 @@
+type: ddflare
+display_name: Datadog Flare Extension
+
+status:
+  class: extension
+  stability:
+    beta: [extension]
+  codeowners:
+    active: [DataDog/opentelemetry]

comp/otelcol/ddprofilingextension/impl/config.go

@@ -9,8 +9,8 @@ package ddprofilingextensionimpl
 // Config contains the config of the profiler.
 type Config struct {
 	ProfilerOptions ProfilerOptions `mapstructure:"profiler_options"`
-	// Endpoint reports the endpoint used for profiles.
-	// Default: BuildInfo.Version (e.g. v0.117.0)
+	// Endpoint is the local port the profiling HTTP server listens on; used as "localhost:<endpoint>".
+	// Default: "7501"
 	Endpoint string `mapstructure:"endpoint"`
 }
 

comp/otelcol/ddprofilingextension/impl/config.schema.yaml

@@ -0,0 +1,30 @@
+$defs:
+  profiler_options:
+    description: ProfilerOptions defines settings relevant to the profiler.
+    type: object
+    properties:
+      env:
+        description: 'Env the profiler will report with. Default: none'
+        type: string
+      period:
+        description: 'Period in seconds the profiler will report with. Default: 60s'
+        type: integer
+      profile_types:
+        description: ProfileTypes specifies additional profile types to enable. supported values are blockprofile, mutexprofile and goroutineprofile. By default CPU and Heap profiles are enabled.
+        type: array
+        items:
+          type: string
+      service:
+        description: 'Service the profiler will report with. Default: BuildInfo.Command (e.g. otel-agent)'
+        type: string
+      version:
+        description: 'Version the profiler will report with. Default: BuildInfo.Version (e.g. v0.117.0)'
+        type: string
+description: Config contains the config of the profiler.
+type: object
+properties:
+  endpoint:
+    description: 'Endpoint is the local port the profiling HTTP server listens on; used as "localhost:<endpoint>". Default: "7501"'
+    type: string
+  profiler_options:
+    $ref: profiler_options

comp/otelcol/ddprofilingextension/impl/metadata.yaml

@@ -0,0 +1,9 @@
+type: ddprofiling
+display_name: Datadog Profiling Extension
+
+status:
+  class: extension
+  stability:
+    development: [extension]
+  codeowners:
+    active: [DataDog/opentelemetry]

comp/otelcol/dogtelextension/impl/config.go

@@ -12,46 +12,67 @@ import (
 
 // Config defines the configuration for the dogtelextension
 type Config struct {
-	// Metadata collection settings.
-	// EnableMetadataCollection is a pointer so that nil ("not set") is
-	// distinguishable from an explicit false, allowing NewConfigComponent to
-	// leave the DD agent default (true) intact when the field is absent.
+	// EnableMetadataCollection toggles metadata collection. Pointer so that
+	// nil ("not set") is distinguishable from an explicit false, allowing
+	// NewConfigComponent to leave the DD agent default (true) intact when
+	// the field is absent.
 	EnableMetadataCollection *bool `mapstructure:"enable_metadata_collection"`
-	MetadataInterval         int   `mapstructure:"metadata_interval"` // seconds; 0 = use agent default (1800)
+	// MetadataInterval is the metadata collection interval, in seconds.
+	// 0 uses the agent default (1800).
+	MetadataInterval int `mapstructure:"metadata_interval"`
 
-	// Tagger server settings
-	EnableTaggerServer      bool   `mapstructure:"enable_tagger_server"`
-	TaggerServerPort        int    `mapstructure:"tagger_server_port"`         // 0 = auto-assign
-	TaggerServerAddr        string `mapstructure:"tagger_server_addr"`         // Default: localhost
-	TaggerMaxMessageSize    int    `mapstructure:"tagger_max_message_size"`    // Default: 4MB
-	TaggerMaxConcurrentSync int    `mapstructure:"tagger_max_concurrent_sync"` // Default: 5
+	// EnableTaggerServer enables the tagger gRPC server.
+	EnableTaggerServer bool `mapstructure:"enable_tagger_server"`
+	// TaggerServerPort is the port the tagger server binds to. 0 auto-assigns.
+	TaggerServerPort int `mapstructure:"tagger_server_port"`
+	// TaggerServerAddr is the address the tagger server binds to. Default: localhost.
+	TaggerServerAddr string `mapstructure:"tagger_server_addr"`
+	// TaggerMaxMessageSize is the maximum gRPC message size in bytes. Default: 4MB.
+	TaggerMaxMessageSize int `mapstructure:"tagger_max_message_size"`
+	// TaggerMaxConcurrentSync is the maximum number of concurrent tag stream
+	// subscribers. Default: 5.
+	TaggerMaxConcurrentSync int `mapstructure:"tagger_max_concurrent_sync"`
 
-	// Standalone mode (controls secrets and other features)
-	// This is typically set via DD_OTEL_STANDALONE environment variable
+	// StandaloneMode controls whether the extension runs in standalone mode,
+	// enabling secrets resolution and other agent features. Typically set via
+	// the DD_OTEL_STANDALONE environment variable.
 	StandaloneMode bool `mapstructure:"standalone_mode"`
 
 	// Hostname overrides the agent hostname used in standalone mode.
-	// Maps to the "hostname" DD agent config key.
-	// When empty the agent resolves the hostname automatically.
+	// Maps to the "hostname" DD agent config key. When empty the agent
+	// resolves the hostname automatically.
 	Hostname string `mapstructure:"hostname"`
 
-	// Secrets backend settings (standalone mode).
-	// These map directly to the corresponding DD agent config keys so that
-	// ENC[] handles in the OTel config can be resolved without a separate
-	// datadog.yaml file.
-	SecretBackendCommand       string   `mapstructure:"secret_backend_command"`         // path to the resolver binary
-	SecretBackendArguments     []string `mapstructure:"secret_backend_arguments"`       // extra CLI arguments
-	SecretBackendTimeout       int      `mapstructure:"secret_backend_timeout"`         // seconds; 0 = agent default (30s)
-	SecretBackendOutputMaxSize int      `mapstructure:"secret_backend_output_max_size"` // bytes; 0 = agent default
+	// SecretBackendCommand is the path to the secret resolver binary
+	// (standalone mode). Maps to the "secret_backend_command" DD agent
+	// config key so ENC[] handles in the OTel config can be resolved
+	// without a separate datadog.yaml file.
+	SecretBackendCommand string `mapstructure:"secret_backend_command"`
+	// SecretBackendArguments are extra CLI arguments passed to the secret
+	// resolver binary.
+	SecretBackendArguments []string `mapstructure:"secret_backend_arguments"`
+	// SecretBackendTimeout is the secret-resolver execution timeout, in
+	// seconds. 0 uses the agent default (30s).
+	SecretBackendTimeout int `mapstructure:"secret_backend_timeout"`
+	// SecretBackendOutputMaxSize is the maximum secret-resolver output size,
+	// in bytes. 0 uses the agent default.
+	SecretBackendOutputMaxSize int `mapstructure:"secret_backend_output_max_size"`
 
-	// Kubernetes / kubelet settings for K8s tag enrichment (standalone mode).
-	// When the otel-agent runs on a Kubernetes node these allow the workloadmeta
-	// kubelet collector and the local tagger to reach the kubelet API without
+	// KubernetesKubeletHost is the kubelet host used for K8s tag enrichment
+	// in standalone mode (e.g. "status.hostIP" or an explicit IP). When the
+	// otel-agent runs on a Kubernetes node these settings let the workloadmeta
+	// kubelet collector and the local tagger reach the kubelet API without
 	// needing a separate datadog.yaml.
-	KubernetesKubeletHost      string `mapstructure:"kubernetes_kubelet_host"`       // e.g. "status.hostIP" or an explicit IP
-	KubeletTLSVerify           *bool  `mapstructure:"kubelet_tls_verify"`            // nil = keep DD agent default (true)
-	KubernetesHTTPKubeletPort  int    `mapstructure:"kubernetes_http_kubelet_port"`  // 0 = agent default (10255)
-	KubernetesHTTPSKubeletPort int    `mapstructure:"kubernetes_https_kubelet_port"` // 0 = agent default (10250)
+	KubernetesKubeletHost string `mapstructure:"kubernetes_kubelet_host"`
+	// KubeletTLSVerify controls TLS verification for the kubelet client.
+	// nil keeps the DD agent default (true).
+	KubeletTLSVerify *bool `mapstructure:"kubelet_tls_verify"`
+	// KubernetesHTTPKubeletPort is the kubelet HTTP port. 0 uses the agent
+	// default (10255).
+	KubernetesHTTPKubeletPort int `mapstructure:"kubernetes_http_kubelet_port"`
+	// KubernetesHTTPSKubeletPort is the kubelet HTTPS port. 0 uses the agent
+	// default (10250).
+	KubernetesHTTPSKubeletPort int `mapstructure:"kubernetes_https_kubelet_port"`
 }
 
 // Validate validates the configuration

comp/otelcol/dogtelextension/impl/config.schema.yaml

@@ -0,0 +1,58 @@
+description: Config defines the configuration for the dogtelextension
+type: object
+properties:
+  enable_metadata_collection:
+    description: EnableMetadataCollection toggles metadata collection. Pointer so that nil ("not set") is distinguishable from an explicit false, allowing NewConfigComponent to leave the DD agent default (true) intact when the field is absent.
+    x-pointer: true
+    type: boolean
+  enable_tagger_server:
+    description: EnableTaggerServer enables the tagger gRPC server.
+    type: boolean
+  hostname:
+    description: Hostname overrides the agent hostname used in standalone mode. Maps to the "hostname" DD agent config key. When empty the agent resolves the hostname automatically.
+    type: string
+  kubelet_tls_verify:
+    description: KubeletTLSVerify controls TLS verification for the kubelet client. nil keeps the DD agent default (true).
+    x-pointer: true
+    type: boolean
+  kubernetes_http_kubelet_port:
+    description: KubernetesHTTPKubeletPort is the kubelet HTTP port. 0 uses the agent default (10255).
+    type: integer
+  kubernetes_https_kubelet_port:
+    description: KubernetesHTTPSKubeletPort is the kubelet HTTPS port. 0 uses the agent default (10250).
+    type: integer
+  kubernetes_kubelet_host:
+    description: KubernetesKubeletHost is the kubelet host used for K8s tag enrichment in standalone mode (e.g. "status.hostIP" or an explicit IP). When the otel-agent runs on a Kubernetes node these settings let the workloadmeta kubelet collector and the local tagger reach the kubelet API without needing a separate datadog.yaml.
+    type: string
+  metadata_interval:
+    description: MetadataInterval is the metadata collection interval, in seconds. 0 uses the agent default (1800).
+    type: integer
+  secret_backend_arguments:
+    description: SecretBackendArguments are extra CLI arguments passed to the secret resolver binary.
+    type: array
+    items:
+      type: string
+  secret_backend_command:
+    description: SecretBackendCommand is the path to the secret resolver binary (standalone mode). Maps to the "secret_backend_command" DD agent config key so ENC[] handles in the OTel config can be resolved without a separate datadog.yaml file.
+    type: string
+  secret_backend_output_max_size:
+    description: SecretBackendOutputMaxSize is the maximum secret-resolver output size, in bytes. 0 uses the agent default.
+    type: integer
+  secret_backend_timeout:
+    description: SecretBackendTimeout is the secret-resolver execution timeout, in seconds. 0 uses the agent default (30s).
+    type: integer
+  standalone_mode:
+    description: StandaloneMode controls whether the extension runs in standalone mode, enabling secrets resolution and other agent features. Typically set via the DD_OTEL_STANDALONE environment variable.
+    type: boolean
+  tagger_max_concurrent_sync:
+    description: 'TaggerMaxConcurrentSync is the maximum number of concurrent tag stream subscribers. Default: 5.'
+    type: integer
+  tagger_max_message_size:
+    description: 'TaggerMaxMessageSize is the maximum gRPC message size in bytes. Default: 4MB.'
+    type: integer
+  tagger_server_addr:
+    description: 'TaggerServerAddr is the address the tagger server binds to. Default: localhost.'
+    type: string
+  tagger_server_port:
+    description: TaggerServerPort is the port the tagger server binds to. 0 auto-assigns.
+    type: integer

comp/otelcol/dogtelextension/impl/metadata.yaml

@@ -0,0 +1,9 @@
+type: dogtel
+display_name: Datadog Telemetry Extension
+
+status:
+  class: extension
+  stability:
+    beta: [extension]
+  codeowners:
+    active: [DataDog/opentelemetry]

comp/otelcol/otlp/components/exporter/datadogexporter/README.md

@@ -0,0 +1,20 @@
+# Datadog Exporter
+
+This package provides the Datadog exporter factory used inside the Datadog
+Distribution of OpenTelemetry (DDOT) Collector. It builds on top of the
+[`serializerexporter`](../serializerexporter), [`logsagentexporter`](../logsagentexporter),
+and the trace agent component to send metrics, logs, and traces (plus APM
+stats) to Datadog using the agent's internal pipelines.
+
+## Configuration
+
+This factory does not define its own configuration struct. It accepts the
+same configuration as the upstream Datadog exporter in
+[`opentelemetry-collector-contrib`](https://github.com/open-telemetry/opentelemetry-collector-contrib),
+which is the `Config` type from
+[`pkg/datadog/config`](https://github.com/open-telemetry/opentelemetry-collector-contrib/tree/main/pkg/datadog/config).
+
+The corresponding configuration schema is published upstream at
+[`pkg/datadog/config/config.schema.yaml`](https://github.com/open-telemetry/opentelemetry-collector-contrib/blob/main/pkg/datadog/config/config.schema.yaml).
+That single file is the source of truth for every option this exporter
+accepts (API, host metadata, metrics, traces, logs, retry, queue, etc.).

comp/otelcol/otlp/components/exporter/logsagentexporter/config.schema.yaml

@@ -0,0 +1,9 @@
+description: Config defines configuration for the logs agent exporter.
+type: object
+properties:
+  host_metadata:
+    description: HostMetadata defines the host metadata specific configuration
+    $ref: github.com/open-telemetry/opentelemetry-collector-contrib/pkg/datadog/config.host_metadata_config
+  sending_queue:
+    x-optional: true
+    $ref: go.opentelemetry.io/collector/exporter/exporterhelper.queue_batch_config

comp/otelcol/otlp/components/exporter/logsagentexporter/factory.go

@@ -39,14 +39,14 @@ const (
 
 // Config defines configuration for the logs agent exporter.
 type Config struct {
-	OtelSource    string
-	LogSourceName string
+	OtelSource    string                                                   `mapstructure:"-"`
+	LogSourceName string                                                   `mapstructure:"-"`
 	QueueSettings configoptional.Optional[exporterhelper.QueueBatchConfig] `mapstructure:"sending_queue"`
 
 	// HostMetadata defines the host metadata specific configuration
 	HostMetadata datadogconfig.HostMetadataConfig `mapstructure:"host_metadata"`
 
-	OrchestratorConfig OrchestratorConfig
+	OrchestratorConfig OrchestratorConfig `mapstructure:"-"`
 }
 
 // OrchestratorConfig contains configuration for sending orchestrator data to Datadog.

comp/otelcol/otlp/components/exporter/logsagentexporter/metadata.yaml

@@ -0,0 +1,9 @@
+type: logsagent
+display_name: Datadog Logs Agent Exporter
+
+status:
+  class: exporter
+  stability:
+    stable: [logs]
+  codeowners:
+    active: [DataDog/opentelemetry]

comp/otelcol/otlp/components/exporter/serializerexporter/config.schema.yaml

@@ -0,0 +1,32 @@
+$defs:
+  metrics_config:
+    description: MetricsConfig defines the metrics exporter specific configuration options
+    type: object
+    properties:
+      apm_stats_receiver_addr:
+        description: APMStatsReceiverAddr is the address to send APM stats to.
+        type: string
+      tags:
+        description: Tags is a comma-separated list of tags to add to all metrics.
+        type: string
+    allOf:
+      - $ref: github.com/open-telemetry/opentelemetry-collector-contrib/pkg/datadog/config.metrics_config
+description: ExporterConfig defines configuration for the serializer exporter.
+type: object
+properties:
+  api:
+    description: API defines the Datadog API configuration. It is useful for OSS OpenTelemetry Collector or to use the serializer exporter with the OCB.
+    $ref: github.com/open-telemetry/opentelemetry-collector-contrib/pkg/datadog/config.api_config
+  host_metadata:
+    description: HostMetadataConfig defines the host metadata related configuration.
+    $ref: github.com/open-telemetry/opentelemetry-collector-contrib/pkg/datadog/config.host_metadata_config
+  metrics:
+    $ref: metrics_config
+  sending_queue:
+    x-optional: true
+    $ref: go.opentelemetry.io/collector/exporter/exporterhelper.queue_batch_config
+  tls:
+    $ref: go.opentelemetry.io/collector/config/configtls.client_config
+allOf:
+  - $ref: go.opentelemetry.io/collector/exporter/exporterhelper.timeout_config
+  - $ref: go.opentelemetry.io/collector/config/confighttp.client_config

comp/otelcol/otlp/components/exporter/serializerexporter/metadata.yaml

@@ -0,0 +1,9 @@
+type: serializer
+display_name: Datadog Serializer Exporter
+
+status:
+  class: exporter
+  stability:
+    stable: [metrics]
+  codeowners:
+    active: [DataDog/opentelemetry]

comp/otelcol/otlp/components/processor/infraattributesprocessor/config.go

@@ -13,6 +13,9 @@ import (
 
 // Config defines configuration for processor.
 type Config struct {
+	// Cardinality controls which tag cardinality is enriched onto the signal.
+	// Accepted values: 0 = low (host-level, default), 1 = orchestrator
+	// (per-pod/task), 2 = high (per-container/request).
 	Cardinality           types.TagCardinality `mapstructure:"cardinality"`
 	AllowHostnameOverride bool                 `mapstructure:"allow_hostname_override"`
 }