chore: update helm-charts for OTEL trace / metrics headers changes

helm-charts/bifrost/README.md

@@ -8,6 +8,11 @@ Official Helm charts for deploying [Bifrost](https://github.com/maximhq/bifrost)
 
 ## Changelog
 
+### Upcoming
+
+- Added `traces_enabled` to OTEL config (single-profile and `profiles[*]`). When `false`, no traces are exported and `collector_url` is not required, giving a metrics-only profile. Renders into `traces_enabled`; defaults to `true`.
+- Added `trace_headers` and `metrics_headers` to OTEL config (single-profile and `profiles[*]`). Common `headers` still go to both endpoints; these are overlaid on top for the trace / metrics endpoint respectively (same key wins), e.g. a Databricks table name required only on metrics. Render into `trace_headers` / `metrics_headers`.
+
 ### 2.1.33
 
 - Fixed Helm schema validation failure for multi-profile OTEL configs (`bifrost.plugins.otel.config.profiles`), introduced by the `export_timeout` default in 2.1.32.

helm-charts/bifrost/templates/_helpers.tpl

@@ -1372,6 +1372,9 @@ false
 {{- if $inputConfig.service_name }}
 {{- $_ := set $otelConfig "service_name" $inputConfig.service_name }}
 {{- end }}
+{{- if hasKey $inputConfig "traces_enabled" }}
+{{- $_ := set $otelConfig "traces_enabled" $inputConfig.traces_enabled }}
+{{- end }}
 {{- if $inputConfig.collector_url }}
 {{- $_ := set $otelConfig "collector_url" $inputConfig.collector_url }}
 {{- end }}
@@ -1396,6 +1399,12 @@ false
 {{- if $inputConfig.headers }}
 {{- $_ := set $otelConfig "headers" $inputConfig.headers }}
 {{- end }}
+{{- if $inputConfig.trace_headers }}
+{{- $_ := set $otelConfig "trace_headers" $inputConfig.trace_headers }}
+{{- end }}
+{{- if $inputConfig.metrics_headers }}
+{{- $_ := set $otelConfig "metrics_headers" $inputConfig.metrics_headers }}
+{{- end }}
 {{- if $inputConfig.tls_ca_cert }}
 {{- $_ := set $otelConfig "tls_ca_cert" $inputConfig.tls_ca_cert }}
 {{- end }}
@@ -1892,13 +1901,19 @@ Call this template at the beginning of deployment/stateful templates
 {{- $profileEnabled = $profile.enabled }}
 {{- end }}
 {{- if $profileEnabled }}
+{{- $tracesEnabled := true }}
+{{- if hasKey $profile "traces_enabled" }}
+{{- $tracesEnabled = $profile.traces_enabled }}
+{{- end }}
+{{- if $tracesEnabled }}
 {{- if not $profile.collector_url }}
-{{- fail (printf "ERROR: bifrost.plugins.otel.config.profiles[%d].collector_url is required for enabled OTEL profiles." $idx) }}
+{{- fail (printf "ERROR: bifrost.plugins.otel.config.profiles[%d].collector_url is required when traces_enabled is true." $idx) }}
 {{- end }}
 {{- if not $profile.trace_type }}
-{{- fail (printf "ERROR: bifrost.plugins.otel.config.profiles[%d].trace_type is required. Supported values: genai_extension, vercel, open_inference" $idx) }}
+{{- fail (printf "ERROR: bifrost.plugins.otel.config.profiles[%d].trace_type is required when traces_enabled is true. Supported values: genai_extension, vercel, open_inference" $idx) }}
 {{- end }}
-{{- if not $profile.protocol }}
+{{- end }}
+{{- if and (or $tracesEnabled $profile.metrics_enabled) (not $profile.protocol) }}
 {{- fail (printf "ERROR: bifrost.plugins.otel.config.profiles[%d].protocol is required. Supported values: http, grpc" $idx) }}
 {{- end }}
 {{- if and $profile.metrics_enabled (not $profile.metrics_endpoint) }}
@@ -1907,14 +1922,20 @@ Call this template at the beginning of deployment/stateful templates
 {{- end }}
 {{- end }}
 {{- else }}
+{{- $tracesEnabled := true }}
+{{- if hasKey $otelInputConfig "traces_enabled" }}
+{{- $tracesEnabled = $otelInputConfig.traces_enabled }}
+{{- end }}
+{{- if $tracesEnabled }}
 {{- if not $otelInputConfig.collector_url }}
-{{- fail "ERROR: bifrost.plugins.otel.config.collector_url is required when OTEL plugin is enabled. Provide the URL of your OpenTelemetry collector." }}
+{{- fail "ERROR: bifrost.plugins.otel.config.collector_url is required when traces_enabled is true. Provide the URL of your OpenTelemetry collector." }}
 {{- end }}
 {{- if not $otelInputConfig.trace_type }}
-{{- fail "ERROR: bifrost.plugins.otel.config.trace_type is required when OTEL plugin is enabled. Supported values: genai_extension, vercel, open_inference" }}
+{{- fail "ERROR: bifrost.plugins.otel.config.trace_type is required when traces_enabled is true. Supported values: genai_extension, vercel, open_inference" }}
+{{- end }}
 {{- end }}
-{{- if not $otelInputConfig.protocol }}
-{{- fail "ERROR: bifrost.plugins.otel.config.protocol is required when OTEL plugin is enabled. Supported values: http, grpc" }}
+{{- if and (or $tracesEnabled $otelInputConfig.metrics_enabled) (not $otelInputConfig.protocol) }}
+{{- fail "ERROR: bifrost.plugins.otel.config.protocol is required. Supported values: http, grpc" }}
 {{- end }}
 {{- if and $otelInputConfig.metrics_enabled (not $otelInputConfig.metrics_endpoint) }}
 {{- fail "ERROR: bifrost.plugins.otel.config.metrics_endpoint is required when metrics_enabled is true." }}

helm-charts/bifrost/values.schema.json

@@ -1059,11 +1059,12 @@
                   "type": "object",
                   "properties": {
                     "service_name": {
-                      "type": "string"
+                      "type": "string",
+                      "description": "Name of the service to report to Datadog. Supports env.VAR_NAME prefix for environment variable substitution (e.g. env.BIFROST_DD_SERVICE)"
                     },
                     "ml_app": {
                       "type": "string",
-                      "description": "ML application name for Datadog LLM Observability grouping (defaults to service_name)"
+                      "description": "ML application name for Datadog LLM Observability grouping (defaults to service_name). Supports env.VAR_NAME prefix for environment variable substitution (e.g. env.BIFROST_DD_ML_APP)"
                     },
                     "agent_addr": {
                       "type": "string",
@@ -1090,10 +1091,12 @@
                       "description": "DogStatsD server port for metrics, used with dogstatsd_host (agent mode only). Supports env.VAR_NAME prefix. Defaults to 8125; has no effect when dogstatsd_host is unset"
                     },
                     "env": {
-                      "type": "string"
+                      "type": "string",
+                      "description": "Environment tag (e.g. production, staging). Supports env.VAR_NAME prefix for environment variable substitution (e.g. env.BIFROST_DD_ENV)"
                     },
                     "version": {
-                      "type": "string"
+                      "type": "string",
+                      "description": "Service version tag. Supports env.VAR_NAME prefix for environment variable substitution (e.g. env.BIFROST_DD_VERSION)"
                     },
                     "custom_tags": {
                       "type": "object",
@@ -4897,6 +4900,11 @@
           "description": "Whether this profile exports traces and metrics",
           "default": true
         },
+        "traces_enabled": {
+          "type": "boolean",
+          "description": "Enable trace export for this profile. When false, no traces are sent and collector_url is not required, giving a metrics-only profile (pair with metrics_enabled).",
+          "default": true
+        },
         "service_name": {
           "type": "string",
           "description": "Service name to be used for tracing",
@@ -4944,7 +4952,21 @@
           "additionalProperties": {
             "type": "string"
           },
-          "description": "Custom headers for the collector (supports env.VAR_NAME prefix)"
+          "description": "Custom headers sent to both the trace and metrics endpoints (supports env.VAR_NAME prefix)"
+        },
+        "trace_headers": {
+          "type": "object",
+          "additionalProperties": {
+            "type": "string"
+          },
+          "description": "Extra headers sent only to the trace endpoint, overlaid on headers (same key wins). Supports env.VAR_NAME prefix."
+        },
+        "metrics_headers": {
+          "type": "object",
+          "additionalProperties": {
+            "type": "string"
+          },
+          "description": "Extra headers sent only to the metrics endpoint, overlaid on headers (same key wins). Useful when a collector requires a per-signal header, e.g. a Databricks table name. Supports env.VAR_NAME prefix."
         },
         "tls_ca_cert": {
           "type": "string",
@@ -4984,30 +5006,58 @@
       "allOf": [
         {
           "if": {
-            "not": {
-              "properties": {
-                "enabled": {
-                  "const": false
+            "allOf": [
+              {
+                "not": {
+                  "properties": {
+                    "enabled": {
+                      "const": false
+                    }
+                  },
+                  "required": ["enabled"]
                 }
               },
-              "required": ["enabled"]
-            }
+              {
+                "not": {
+                  "properties": {
+                    "traces_enabled": {
+                      "const": false
+                    }
+                  },
+                  "required": ["traces_enabled"]
+                }
+              }
+            ]
           },
           "then": {
             "required": ["collector_url", "trace_type", "protocol"]
           }
         },
         {
           "if": {
-            "properties": {
-              "metrics_enabled": {
-                "const": true
+            "allOf": [
+              {
+                "not": {
+                  "properties": {
+                    "enabled": {
+                      "const": false
+                    }
+                  },
+                  "required": ["enabled"]
+                }
+              },
+              {
+                "properties": {
+                  "metrics_enabled": {
+                    "const": true
+                  }
+                },
+                "required": ["metrics_enabled"]
               }
-            },
-            "required": ["metrics_enabled"]
+            ]
           },
           "then": {
-            "required": ["metrics_endpoint"]
+            "required": ["metrics_endpoint", "protocol"]
           }
         }
       ],

helm-charts/bifrost/values.yaml

@@ -612,14 +612,17 @@ bifrost:
         # Multi-profile shape (use profiles OR the flat single-profile fields below, not both):
         # profiles:
         #   - service_name: "bifrost"
+        #     traces_enabled: true    # Set false for a metrics-only profile (collector_url not required)
         #     collector_url: ""       # e.g., http://otel-collector:4318 (HTTP) or otel-collector:4317 (gRPC)
         #     trace_type: "genai_extension" # genai_extension | vercel | open_inference
         #     protocol: "grpc"        # http | grpc
         #     export_timeout: 5       # Max seconds for one trace export (1-60); the only timeout on gRPC exports
         #     metrics_enabled: false
         #     metrics_endpoint: ""    # e.g., http://otel-collector:4318/v1/metrics (HTTP) or otel-collector:4317 (gRPC)
         #     metrics_push_interval: 15 # Push interval in seconds (1-300)
-        #     headers: {}
+        #     headers: {}             # Sent to both endpoints (supports env.VAR_NAME prefix)
+        #     trace_headers: {}       # Sent only to the trace endpoint, on top of headers (same key overrides)
+        #     metrics_headers: {}     # Sent only to the metrics endpoint, on top of headers (e.g. a Databricks table name)
         #     tls_ca_cert: ""         # Path to TLS CA certificate file
         #     insecure: true          # Skip TLS verification (ignored if tls_ca_cert is set)
         #     disable_content_logging: false
@@ -629,6 +632,8 @@ bifrost:
         #
         # Single-profile shape:
         service_name: "bifrost"
+        # Set false for a metrics-only profile (collector_url not required)
+        # traces_enabled: true
         collector_url: "" # e.g., http://otel-collector:4318 (HTTP) or otel-collector:4317 (gRPC)
         trace_type: "genai_extension" # genai_extension | vercel | open_inference
         protocol: "grpc" # http | grpc
@@ -640,8 +645,12 @@ bifrost:
         metrics_enabled: false
         metrics_endpoint: "" # e.g., http://otel-collector:4318/v1/metrics (HTTP) or otel-collector:4317 (gRPC)
         metrics_push_interval: 15 # Push interval in seconds (1-300)
-        # Custom headers for the collector (supports env.VAR_NAME prefix)
+        # Custom headers sent to both the trace and metrics endpoints (supports env.VAR_NAME prefix)
         headers: {}
+        # Extra headers sent only to the trace endpoint, on top of headers (same key overrides)
+        # trace_headers: {}
+        # Extra headers sent only to the metrics endpoint, on top of headers (e.g. a Databricks table name)
+        # metrics_headers: {}
         # TLS configuration
         tls_ca_cert: "" # Path to TLS CA certificate file
         # insecure: false # Skip TLS verification (ignored if tls_ca_cert is set)
@@ -658,7 +667,7 @@ bifrost:
       enabled: false
       version: 1
       config:
-        service_name: "bifrost"
+        service_name: "bifrost"  # Service name reported to Datadog (supports env.VAR_NAME)
         # Datadog Agent address. Supports env.VAR_NAME references — e.g. set
         # agent_addr: "env.DD_AGENT_ADDR" and inject DD_AGENT_ADDR via the
         # top-level `env:` (e.g. from status.hostIP for a node-local agent DaemonSet).
@@ -672,11 +681,11 @@ bifrost:
         # agent_port: "8126"
         # dogstatsd_host: "env.DD_AGENT_HOST"
         # dogstatsd_port: "8125"
-        env: ""
-        version: ""
+        env: ""                     # Environment tag, e.g. production (supports env.VAR_NAME)
+        version: ""                 # Service version tag (supports env.VAR_NAME)
         custom_tags: {}
         enable_traces: true
-        # ml_app: ""                  # ML app name for LLM Observability (defaults to service_name)
+        # ml_app: ""                  # ML app name for LLM Observability (defaults to service_name, supports env.VAR_NAME)
         # enable_metrics: true
         # enable_llm_obs: true
         # group_traces_by_session: false  # Group requests sharing x-bf-session-id into one APM trace (agent mode only)