feat: add metrics query API

llama_stack/apis/telemetry/telemetry.py

@@ -203,6 +203,45 @@ class QuerySpanTreeResponse(BaseModel):
     data: dict[str, SpanWithStatus]
 
 
+@json_schema_type
+class MetricQueryType(Enum):
+    RANGE = "range"
+    INSTANT = "instant"
+
+
+@json_schema_type
+class MetricLabelOperator(Enum):
+    EQUALS = "="
+    NOT_EQUALS = "!="
+    REGEX_MATCH = "=~"
+    REGEX_NOT_MATCH = "!~"
+
+
+@json_schema_type
+class MetricLabelMatcher(BaseModel):
+    name: str
+    value: str
+    operator: MetricLabelOperator = MetricLabelOperator.EQUALS
+
+
+@json_schema_type
+class MetricDataPoint(BaseModel):
+    timestamp: datetime
+    value: float
+
+
+@json_schema_type
+class MetricSeries(BaseModel):
+    metric: str
+    labels: Dict[str, str]
+    values: List[MetricDataPoint]
+
+
+@json_schema_type
+class GetMetricsResponse(BaseModel):
+    data: List[MetricSeries]
+
+
 @runtime_checkable
 class Telemetry(Protocol):
     @webmethod(route="/telemetry/events", method="POST")
@@ -247,3 +286,14 @@ class Telemetry(Protocol):
         dataset_id: str,
         max_depth: int | None = None,
     ) -> None: ...
+
+    @webmethod(route="/telemetry/metrics/{metric_name}", method="POST")
+    async def get_metrics(
+        self,
+        metric_name: str,
+        start_time: int,
+        end_time: Optional[int] = None,
+        step: Optional[str] = "1d",
+        query_type: MetricQueryType = MetricQueryType.RANGE,
+        label_matchers: Optional[List[MetricLabelMatcher]] = None,
+    ) -> GetMetricsResponse: ...

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

@@ -19,7 +19,10 @@ from opentelemetry.semconv.resource import ResourceAttributes
 
 from llama_stack.apis.telemetry import (
     Event,
+    GetMetricsResponse,
     MetricEvent,
+    MetricLabelMatcher,
+    MetricQueryType,
     QueryCondition,
     QuerySpanTreeResponse,
     QueryTracesResponse,
@@ -123,6 +126,17 @@ class TelemetryAdapter(TelemetryDatasetMixin, Telemetry):
         else:
             raise ValueError(f"Unknown event type: {event}")
 
+    async def get_metrics(
+        self,
+        metric_name: str,
+        start_time: int,
+        end_time: Optional[int] = None,
+        step: Optional[str] = "1d",
+        query_type: MetricQueryType = MetricQueryType.RANGE,
+        label_matchers: Optional[List[MetricLabelMatcher]] = None,
+    ) -> GetMetricsResponse:
+        pass
+
     def _log_unstructured(self, event: UnstructuredLogEvent, ttl_seconds: int) -> None:
         with self._lock:
             # Use global storage instead of instance storage