llama-stack-mirror/llama_stack/apis/telemetry/telemetry.py
Dinesh Yeduguru fcd6449519
Telemetry API redesign (#525)
# What does this PR do?
Change the Telemetry API to be able to support different use cases like
returning traces for the UI and ability to export for Evals.
Other changes:
* Add a new trace_protocol decorator to decorate all our API methods so
that any call to them will automatically get traced across all impls.
* There is some issue with the decorator pattern of span creation when
using async generators, where there are multiple yields with in the same
context. I think its much more explicit by using the explicit context
manager pattern using with. I moved the span creations in agent instance
to be using with
* Inject session id at the turn level, which should quickly give us all
traces across turns for a given session

Addresses #509

## Test Plan
```
llama stack run /Users/dineshyv/.llama/distributions/llamastack-together/together-run.yaml
PYTHONPATH=. python -m examples.agents.rag_with_memory_bank localhost 5000


 curl -X POST 'http://localhost:5000/alpha/telemetry/query-traces' \
-H 'Content-Type: application/json' \
-d '{
  "attribute_filters": [
    {
      "key": "session_id",
      "op": "eq",
      "value": "dd667b87-ca4b-4d30-9265-5a0de318fc65" }],
  "limit": 100,
  "offset": 0,
  "order_by": ["start_time"]
}' | jq .
[
  {
    "trace_id": "6902f54b83b4b48be18a6f422b13e16f",
    "root_span_id": "5f37b85543afc15a",
    "start_time": "2024-12-04T08:08:30.501587",
    "end_time": "2024-12-04T08:08:36.026463"
  },
  {
    "trace_id": "92227dac84c0615ed741be393813fb5f",
    "root_span_id": "af7c5bb46665c2c8",
    "start_time": "2024-12-04T08:08:36.031170",
    "end_time": "2024-12-04T08:08:41.693301"
  },
  {
    "trace_id": "7d578a6edac62f204ab479fba82f77b6",
    "root_span_id": "1d935e3362676896",
    "start_time": "2024-12-04T08:08:41.695204",
    "end_time": "2024-12-04T08:08:47.228016"
  },
  {
    "trace_id": "dbd767d76991bc816f9f078907dc9ff2",
    "root_span_id": "f5a7ee76683b9602",
    "start_time": "2024-12-04T08:08:47.234578",
    "end_time": "2024-12-04T08:08:53.189412"
  }
]


curl -X POST 'http://localhost:5000/alpha/telemetry/get-span-tree' \
-H 'Content-Type: application/json' \
-d '{ "span_id" : "6cceb4b48a156913", "max_depth": 2, "attributes_to_return": ["input"] }' | jq .
  % Total    % Received % Xferd  Average Speed   Time    Time     Time  Current
                                 Dload  Upload   Total   Spent    Left  Speed
100   875  100   790  100    85  18462   1986 --:--:-- --:--:-- --:--:-- 20833
{
  "span_id": "6cceb4b48a156913",
  "trace_id": "dafa796f6aaf925f511c04cd7c67fdda",
  "parent_span_id": "892a66d726c7f990",
  "name": "retrieve_rag_context",
  "start_time": "2024-12-04T09:28:21.781995",
  "end_time": "2024-12-04T09:28:21.913352",
  "attributes": {
    "input": [
      "{\"role\":\"system\",\"content\":\"You are a helpful assistant\"}",
      "{\"role\":\"user\",\"content\":\"What are the top 5 topics that were explained in the documentation? Only list succinct bullet points.\",\"context\":null}"
    ]
  },
  "children": [
    {
      "span_id": "1a2df181854064a8",
      "trace_id": "dafa796f6aaf925f511c04cd7c67fdda",
      "parent_span_id": "6cceb4b48a156913",
      "name": "MemoryRouter.query_documents",
      "start_time": "2024-12-04T09:28:21.787620",
      "end_time": "2024-12-04T09:28:21.906512",
      "attributes": {
        "input": null
      },
      "children": [],
      "status": "ok"
    }
  ],
  "status": "ok"
}

```

<img width="1677" alt="Screenshot 2024-12-04 at 9 42 56 AM"
src="https://github.com/user-attachments/assets/4d3cea93-05ce-415a-93d9-4b1628631bf8">
2024-12-04 11:22:45 -08:00

188 lines
4.2 KiB
Python

# Copyright (c) Meta Platforms, Inc. and affiliates.
# All rights reserved.
#
# This source code is licensed under the terms described in the LICENSE file in
# the root directory of this source tree.
from datetime import datetime
from enum import Enum
from typing import (
Any,
Dict,
List,
Literal,
Optional,
Protocol,
runtime_checkable,
Union,
)
from llama_models.schema_utils import json_schema_type, webmethod
from pydantic import BaseModel, Field
from typing_extensions import Annotated
# Add this constant near the top of the file, after the imports
DEFAULT_TTL_DAYS = 7
@json_schema_type
class SpanStatus(Enum):
OK = "ok"
ERROR = "error"
@json_schema_type
class Span(BaseModel):
span_id: str
trace_id: str
parent_span_id: Optional[str] = None
name: str
start_time: datetime
end_time: Optional[datetime] = None
attributes: Optional[Dict[str, Any]] = Field(default_factory=dict)
def set_attribute(self, key: str, value: Any):
if self.attributes is None:
self.attributes = {}
self.attributes[key] = value
@json_schema_type
class Trace(BaseModel):
trace_id: str
root_span_id: str
start_time: datetime
end_time: Optional[datetime] = None
@json_schema_type
class EventType(Enum):
UNSTRUCTURED_LOG = "unstructured_log"
STRUCTURED_LOG = "structured_log"
METRIC = "metric"
@json_schema_type
class LogSeverity(Enum):
VERBOSE = "verbose"
DEBUG = "debug"
INFO = "info"
WARN = "warn"
ERROR = "error"
CRITICAL = "critical"
class EventCommon(BaseModel):
trace_id: str
span_id: str
timestamp: datetime
attributes: Optional[Dict[str, Any]] = Field(default_factory=dict)
@json_schema_type
class UnstructuredLogEvent(EventCommon):
type: Literal[EventType.UNSTRUCTURED_LOG.value] = EventType.UNSTRUCTURED_LOG.value
message: str
severity: LogSeverity
@json_schema_type
class MetricEvent(EventCommon):
type: Literal[EventType.METRIC.value] = EventType.METRIC.value
metric: str # this would be an enum
value: Union[int, float]
unit: str
@json_schema_type
class StructuredLogType(Enum):
SPAN_START = "span_start"
SPAN_END = "span_end"
@json_schema_type
class SpanStartPayload(BaseModel):
type: Literal[StructuredLogType.SPAN_START.value] = (
StructuredLogType.SPAN_START.value
)
name: str
parent_span_id: Optional[str] = None
@json_schema_type
class SpanEndPayload(BaseModel):
type: Literal[StructuredLogType.SPAN_END.value] = StructuredLogType.SPAN_END.value
status: SpanStatus
StructuredLogPayload = Annotated[
Union[
SpanStartPayload,
SpanEndPayload,
],
Field(discriminator="type"),
]
@json_schema_type
class StructuredLogEvent(EventCommon):
type: Literal[EventType.STRUCTURED_LOG.value] = EventType.STRUCTURED_LOG.value
payload: StructuredLogPayload
Event = Annotated[
Union[
UnstructuredLogEvent,
MetricEvent,
StructuredLogEvent,
],
Field(discriminator="type"),
]
@json_schema_type
class EvalTrace(BaseModel):
session_id: str
step: str
input: str
output: str
expected_output: str
@json_schema_type
class SpanWithChildren(Span):
children: List["SpanWithChildren"] = Field(default_factory=list)
status: Optional[SpanStatus] = None
@json_schema_type
class QueryCondition(BaseModel):
key: str
op: Literal["eq", "ne", "gt", "lt"]
value: Any
@runtime_checkable
class Telemetry(Protocol):
@webmethod(route="/telemetry/log-event")
async def log_event(
self, event: Event, ttl_seconds: int = DEFAULT_TTL_DAYS * 86400
) -> None: ...
@webmethod(route="/telemetry/query-traces", method="POST")
async def query_traces(
self,
attribute_filters: Optional[List[QueryCondition]] = None,
limit: Optional[int] = 100,
offset: Optional[int] = 0,
order_by: Optional[List[str]] = None,
) -> List[Trace]: ...
@webmethod(route="/telemetry/get-span-tree", method="POST")
async def get_span_tree(
self,
span_id: str,
attributes_to_return: Optional[List[str]] = None,
max_depth: Optional[int] = None,
) -> SpanWithChildren: ...