feat: Add new compact MetricInResponse type

docs/_static/llama-stack-spec.html

@@ -4630,8 +4630,6 @@
                 },
                 "additionalProperties": false,
                 "required": [
-                    "trace_id",
-                    "span_id",
                     "timestamp",
                     "type",
                     "metric",
@@ -8450,8 +8448,6 @@
                 },
                 "additionalProperties": false,
                 "required": [
-                    "trace_id",
-                    "span_id",
                     "timestamp",
                     "type",
                     "payload"
@@ -8524,8 +8520,6 @@
                 },
                 "additionalProperties": false,
                 "required": [
-                    "trace_id",
-                    "span_id",
                     "timestamp",
                     "type",
                     "message",

docs/_static/llama-stack-spec.yaml

@@ -3149,8 +3149,6 @@ components:
           type: string
       additionalProperties: false
       required:
-        - trace_id
-        - span_id
         - timestamp
         - type
         - metric
@@ -5760,8 +5758,6 @@ components:
           $ref: '#/components/schemas/StructuredLogPayload'
       additionalProperties: false
       required:
-        - trace_id
-        - span_id
         - timestamp
         - type
         - payload
@@ -5804,8 +5800,6 @@ components:
           $ref: '#/components/schemas/LogSeverity'
       additionalProperties: false
       required:
-        - trace_id
-        - span_id
         - timestamp
         - type
         - message

llama_stack/apis/telemetry/telemetry.py

@@ -75,9 +75,9 @@ class LogSeverity(Enum):
 
 
 class EventCommon(BaseModel):
-    trace_id: str
-    span_id: str
-    timestamp: datetime
+    trace_id: Optional[str] = None
+    span_id: Optional[str] = None
+    timestamp: Optional[datetime] = None
     attributes: Optional[Dict[str, Primitive]] = Field(default_factory=dict)
 
 
@@ -93,7 +93,14 @@ class MetricEvent(EventCommon):
     type: Literal[EventType.METRIC.value] = EventType.METRIC.value
     metric: str  # this would be an enum
     value: Union[int, float]
-    unit: str
+    unit: Optional[str] = None
+
+
+@json_schema_type
+class MetricInResponse(BaseModel):
+    metric: str
+    value: Union[int, float]
+    unit: Optional[str] = None
 
 
 # This is a short term solution to allow inference API to return metrics
@@ -117,7 +124,7 @@ class MetricEvent(EventCommon):
 
 
 class MetricResponseMixin(BaseModel):
-    metrics: Optional[List[MetricEvent]] = None
+    metrics: Optional[List[MetricInResponse]] = None
 
 
 @json_schema_type

llama_stack/distribution/routers/routers.py

@@ -48,7 +48,7 @@ from llama_stack.apis.scoring import (
     ScoringFnParams,
 )
 from llama_stack.apis.shields import Shield
-from llama_stack.apis.telemetry import MetricEvent, Telemetry
+from llama_stack.apis.telemetry import MetricEvent, MetricInResponse, Telemetry
 from llama_stack.apis.tools import (
     RAGDocument,
     RAGQueryConfig,
@@ -206,12 +206,12 @@ class InferenceRouter(Inference):
         completion_tokens: int,
         total_tokens: int,
         model: Model,
-    ) -> List[MetricEvent]:
+    ) -> List[MetricInResponse]:
         metrics = self._construct_metrics(prompt_tokens, completion_tokens, total_tokens, model)
         if self.telemetry:
             for metric in metrics:
                 await self.telemetry.log_event(metric)
-        return metrics
+        return [MetricInResponse(metric=metric.metric, value=metric.value) for metric in metrics]
 
     async def _count_tokens(
         self,

llama_stack/providers/inline/telemetry/meta_reference/telemetry.py

@@ -153,20 +153,20 @@ class TelemetryAdapter(TelemetryDatasetMixin, Telemetry):
             else:
                 print(f"Warning: No active span found for span_id {span_id}. Dropping event: {event}")
 
-    def _get_or_create_counter(self, name: str, unit: str) -> metrics.Counter:
+    def _get_or_create_counter(self, name: str, unit: Optional[str] = None) -> metrics.Counter:
         if name not in _GLOBAL_STORAGE["counters"]:
             _GLOBAL_STORAGE["counters"][name] = self.meter.create_counter(
                 name=name,
-                unit=unit,
+                unit=unit or "",
                 description=f"Counter for {name}",
             )
         return _GLOBAL_STORAGE["counters"][name]
 
-    def _get_or_create_gauge(self, name: str, unit: str) -> metrics.ObservableGauge:
+    def _get_or_create_gauge(self, name: str, unit: Optional[str] = None) -> metrics.ObservableGauge:
         if name not in _GLOBAL_STORAGE["gauges"]:
             _GLOBAL_STORAGE["gauges"][name] = self.meter.create_gauge(
                 name=name,
-                unit=unit,
+                unit=unit or "",
                 description=f"Gauge for {name}",
             )
         return _GLOBAL_STORAGE["gauges"][name]
@@ -181,11 +181,11 @@ class TelemetryAdapter(TelemetryDatasetMixin, Telemetry):
             up_down_counter = self._get_or_create_up_down_counter(event.metric, event.unit)
             up_down_counter.add(event.value, attributes=event.attributes)
 
-    def _get_or_create_up_down_counter(self, name: str, unit: str) -> metrics.UpDownCounter:
+    def _get_or_create_up_down_counter(self, name: str, unit: Optional[str] = None) -> metrics.UpDownCounter:
         if name not in _GLOBAL_STORAGE["up_down_counters"]:
             _GLOBAL_STORAGE["up_down_counters"][name] = self.meter.create_up_down_counter(
                 name=name,
-                unit=unit,
+                unit=unit or "",
                 description=f"UpDownCounter for {name}",
             )
         return _GLOBAL_STORAGE["up_down_counters"][name]