feat(observability): Langfuse 전송 — 턴 마무리(사용자 trace)·게이트웨이(FabriX generation) 두 훅, SDK 없이 HTTP. deploy/langfuse 에 compose+env+반입 절차
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
This commit is contained in:
lee-hyeon-cheol
co-authored by
Claude Fable 5.1
Claude Fable 5.1
parent
95b169a7b9
commit
0321050053
@@ -62,3 +62,8 @@ SAP_API_BASE_ICF=
|
||||
# ── ABAP 소스 인덱스(ABAP_INDEXING, 별도 서버 :8100) — OpenCode 툴 abap_index_* 가 씀 ──
|
||||
# 비우면 기본 127.0.0.1:8100. 서버가 없어도 앱은 돌아감(툴이 "없음"으로 답하고 생성 코드로).
|
||||
ABAP_INDEX_URL=http://127.0.0.1:8100
|
||||
|
||||
# ── Langfuse(LLM 관측, 자체 호스팅 deploy/langfuse/) — HOST 비우면 안 보냄 ──
|
||||
LANGFUSE_HOST=
|
||||
LANGFUSE_PUBLIC_KEY=
|
||||
LANGFUSE_SECRET_KEY=
|
||||
|
||||
@@ -20,12 +20,14 @@ import asyncio
|
||||
import json
|
||||
import logging
|
||||
import time
|
||||
from datetime import datetime, timezone
|
||||
|
||||
from django.conf import settings
|
||||
from django.http import StreamingHttpResponse
|
||||
from django.views.decorators.csrf import csrf_exempt
|
||||
|
||||
from apps.accounts.authentication import user_from_token
|
||||
from apps.gateway import langfuse
|
||||
from asgiref.sync import sync_to_async
|
||||
from common.envelope import json_error
|
||||
from common.opencode_service import opencode_service
|
||||
@@ -248,6 +250,25 @@ async def _finalize(session: ChatSession, state: TurnState, started: float, *, f
|
||||
if title:
|
||||
session.title_llm = title[:200]
|
||||
await session.asave(update_fields=["is_generating", "title_llm", "updated_at"])
|
||||
|
||||
# 관측: 사용자 단위 trace 하나 + 답변 generation 하나. 실패해도 여기까진 이미 저장됨.
|
||||
if langfuse.enabled():
|
||||
tid = f"{session.id}-{int(started * 1000)}"
|
||||
user_email = await sync_to_async(lambda: session.user.email)()
|
||||
meta = {"failed": failed} if failed else {}
|
||||
langfuse.send_later([
|
||||
langfuse.trace(tid, "chat", userId=user_email, sessionId=session.id, input=state.user_text,
|
||||
output=content, metadata=meta, tags=["codeassist"]),
|
||||
langfuse.generation(
|
||||
tid, "opencode-turn",
|
||||
startTime=datetime.fromtimestamp(started, timezone.utc).isoformat(),
|
||||
endTime=langfuse.now_iso(),
|
||||
input=state.user_text, output=content,
|
||||
usage=langfuse.usage_of(usage["input"], usage["output"]) if usage else None,
|
||||
level="ERROR" if failed else "DEFAULT", statusMessage=failed or "",
|
||||
metadata={"elapsed_ms": usage["elapsed_ms"]} if usage else {},
|
||||
),
|
||||
])
|
||||
return usage, title
|
||||
|
||||
|
||||
|
||||
@@ -0,0 +1,85 @@
|
||||
"""Langfuse 로 trace/generation 쏘기 — SDK 없이 HTTP 한 방 (docs-lib/langfuse.md).
|
||||
|
||||
LANGFUSE_HOST 가 비어 있으면 전부 no-op. 보내는 건 fire-and-forget: 실패해도 채팅엔 영향 0, 로그만.
|
||||
두 군데서 부름:
|
||||
- apps/chat/stream.py _finalize → 사용자 단위 trace(누가·어느 세션·질문·최종 답·토큰·시간)
|
||||
- apps/gateway/views.py → FabriX 호출 단위 generation(모델·프롬프트 원문·응답·상태)
|
||||
둘은 서로 모름(OpenCode 가 사이에 있어 사용자 정보가 게이트웨이까지 안 옴). 나중에 잇고 싶으면 sessionId 로 대조.
|
||||
"""
|
||||
|
||||
from __future__ import annotations
|
||||
|
||||
import asyncio
|
||||
import logging
|
||||
import uuid
|
||||
from datetime import datetime, timezone
|
||||
|
||||
import httpx
|
||||
from django.conf import settings
|
||||
|
||||
log = logging.getLogger(__name__)
|
||||
|
||||
TRANSPORT: httpx.AsyncBaseTransport | None = None # 테스트가 MockTransport 꽂는 자리
|
||||
_pending: set[asyncio.Task] = set() # GC 에 안 먹히게 잡아둠
|
||||
|
||||
|
||||
def enabled() -> bool:
|
||||
return bool(settings.LANGFUSE.get("host"))
|
||||
|
||||
|
||||
def now_iso() -> str:
|
||||
return datetime.now(timezone.utc).isoformat().replace("+00:00", "Z")
|
||||
|
||||
|
||||
def _event(kind: str, body: dict) -> dict:
|
||||
return {"id": str(uuid.uuid4()), "timestamp": now_iso(), "type": kind, "body": body}
|
||||
|
||||
|
||||
def trace(trace_id: str, name: str, **body) -> dict:
|
||||
return _event("trace-create", {"id": trace_id, "name": name, **body})
|
||||
|
||||
|
||||
def generation(trace_id: str, name: str, **body) -> dict:
|
||||
return _event("generation-create", {"id": str(uuid.uuid4()), "traceId": trace_id, "name": name, **body})
|
||||
|
||||
|
||||
def usage_of(inp: int | None, out: int | None) -> dict | None:
|
||||
if inp is None and out is None:
|
||||
return None
|
||||
return {"input": inp or 0, "output": out or 0, "total": (inp or 0) + (out or 0)}
|
||||
|
||||
|
||||
async def send(events: list[dict]) -> bool:
|
||||
"""배치 하나 전송. 207 이면 성공(부분 실패는 로그)."""
|
||||
cfg = settings.LANGFUSE
|
||||
if not cfg.get("host") or not events:
|
||||
return False
|
||||
try:
|
||||
async with httpx.AsyncClient(timeout=5.0, transport=TRANSPORT) as client:
|
||||
resp = await client.post(
|
||||
cfg["host"].rstrip("/") + "/api/public/ingestion",
|
||||
json={"batch": events},
|
||||
auth=(cfg.get("public_key", ""), cfg.get("secret_key", "")),
|
||||
)
|
||||
if resp.status_code not in (200, 207):
|
||||
log.warning("langfuse ← %s %s", resp.status_code, resp.text[:200])
|
||||
return False
|
||||
errors = (resp.json() or {}).get("errors") or []
|
||||
if errors:
|
||||
log.warning("langfuse 일부 실패: %s", errors[:3])
|
||||
return not errors
|
||||
except Exception as e: # noqa: BLE001 — 관측용이라 절대 본 흐름 안 깨뜨림
|
||||
log.warning("langfuse 전송 실패: %s: %s", type(e).__name__, e)
|
||||
return False
|
||||
|
||||
|
||||
def send_later(events: list[dict]) -> None:
|
||||
"""지금 흐름 안 막고 백그라운드로. 이벤트 루프 없으면(동기 테스트) 조용히 버림."""
|
||||
if not enabled() or not events:
|
||||
return
|
||||
try:
|
||||
task = asyncio.get_running_loop().create_task(send(events))
|
||||
except RuntimeError:
|
||||
return
|
||||
_pending.add(task)
|
||||
task.add_done_callback(_pending.discard)
|
||||
@@ -11,6 +11,9 @@ from __future__ import annotations
|
||||
|
||||
import asyncio
|
||||
import json
|
||||
import time
|
||||
import uuid
|
||||
from datetime import datetime, timezone
|
||||
import logging
|
||||
from typing import AsyncIterator
|
||||
|
||||
@@ -19,6 +22,7 @@ from django.conf import settings
|
||||
from django.http import HttpRequest, HttpResponse, JsonResponse, StreamingHttpResponse
|
||||
from django.views.decorators.csrf import csrf_exempt
|
||||
|
||||
from . import langfuse
|
||||
from .fabrix import FabrixConfig, Variant, auth_variants
|
||||
|
||||
log = logging.getLogger(__name__)
|
||||
@@ -60,6 +64,57 @@ def _sse_error(status: int, detail: bytes) -> bytes:
|
||||
return b"data: " + json.dumps(payload, ensure_ascii=False).encode() + b"\n\ndata: [DONE]\n\n"
|
||||
|
||||
|
||||
class _Collect:
|
||||
"""상류 응답에서 답변 텍스트·usage 만 긁어 모음(관측용). 스트림은 줄 단위 `data: {json}`."""
|
||||
|
||||
def __init__(self) -> None:
|
||||
self.text: list[str] = []
|
||||
self.usage: dict | None = None
|
||||
self.buf = b""
|
||||
|
||||
def feed(self, chunk: bytes) -> None:
|
||||
self.buf += chunk
|
||||
while b"\n" in self.buf:
|
||||
line, self.buf = self.buf.split(b"\n", 1)
|
||||
self._line(line.strip())
|
||||
|
||||
def _line(self, line: bytes) -> None:
|
||||
if not line.startswith(b"data:") or line.endswith(b"[DONE]"):
|
||||
return
|
||||
try:
|
||||
d = json.loads(line[5:])
|
||||
except ValueError:
|
||||
return
|
||||
self.json(d)
|
||||
|
||||
def json(self, d: dict) -> None:
|
||||
for c in d.get("choices") or []:
|
||||
t = (c.get("delta") or c.get("message") or {}).get("content")
|
||||
if isinstance(t, str):
|
||||
self.text.append(t)
|
||||
if d.get("usage"):
|
||||
self.usage = d["usage"]
|
||||
|
||||
|
||||
def _observe(model_id: str, body: dict, col: _Collect, started: float, status: int, *, kinds: list[str]) -> None:
|
||||
if not langfuse.enabled():
|
||||
return
|
||||
u = col.usage or {}
|
||||
tid = str(uuid.uuid4())
|
||||
langfuse.send_later([
|
||||
langfuse.trace(tid, "fabrix", tags=["gateway"], metadata={"parts": kinds}),
|
||||
langfuse.generation(
|
||||
tid, "fabrix.chat", model=model_id,
|
||||
startTime=datetime.fromtimestamp(started, timezone.utc).isoformat(), endTime=langfuse.now_iso(),
|
||||
input=body.get("messages"), output="".join(col.text),
|
||||
usage=langfuse.usage_of(u.get("prompt_tokens"), u.get("completion_tokens")),
|
||||
level="ERROR" if status >= 400 else "DEFAULT", statusMessage="" if status < 400 else f"upstream {status}",
|
||||
modelParameters={k: v for k, v in body.items() if k in ("temperature", "max_completion_tokens", "max_tokens")},
|
||||
metadata={"stream": bool(body.get("stream")), "tools": len(body.get("tools") or [])},
|
||||
),
|
||||
])
|
||||
|
||||
|
||||
async def healthcheck(_request: HttpRequest) -> JsonResponse:
|
||||
return JsonResponse({"success": True})
|
||||
|
||||
@@ -95,6 +150,9 @@ async def chat_completions(request: HttpRequest) -> HttpResponse:
|
||||
# FabriX 가 맞는 조합에도 가끔 401 을 뱉음(2026-09-21 고객사 실측: curl 10번 중 1~2번).
|
||||
# 4개 조합 다 돌고 나서 제일 유력한 조합(맨 앞)을 잠깐 쉬고 한 번 더 — 그래도 401 이면 진짜 인증 문제
|
||||
variants = variants + variants[:1]
|
||||
model_id = headers.get("x-llm-model-id", "")
|
||||
started = time.time()
|
||||
col = _Collect()
|
||||
|
||||
if body.get("stream"):
|
||||
|
||||
@@ -119,9 +177,11 @@ async def chat_completions(request: HttpRequest) -> HttpResponse:
|
||||
return
|
||||
_last_ok["variant"] = vkey
|
||||
async for chunk in resp.aiter_raw():
|
||||
col.feed(chunk)
|
||||
yield chunk
|
||||
finally:
|
||||
await resp.aclose()
|
||||
_observe(model_id, body, col, started, resp.status_code, kinds=kinds or ["text"])
|
||||
|
||||
out = StreamingHttpResponse(gen(), content_type="text/event-stream")
|
||||
out["Cache-Control"] = "no-cache"
|
||||
@@ -141,4 +201,10 @@ async def chat_completions(request: HttpRequest) -> HttpResponse:
|
||||
break
|
||||
except httpx.HTTPError as e:
|
||||
return _err(502, f"FabriX 호출 실패: {type(e).__name__}: {e}", "upstream_error")
|
||||
if resp.status_code < 400:
|
||||
try:
|
||||
col.json(resp.json())
|
||||
except ValueError:
|
||||
pass
|
||||
_observe(model_id, body, col, started, resp.status_code, kinds=kinds or ["text"])
|
||||
return HttpResponse(resp.content, status=resp.status_code, content_type="application/json")
|
||||
|
||||
@@ -88,6 +88,12 @@ STREAM_FIRST_EVENT_TIMEOUT_S = int(_env("STREAM_FIRST_EVENT_TIMEOUT_S", "300"))
|
||||
from apps.gateway.fabrix import ENV_KEYS as _FABRIX_KEYS # noqa: E402
|
||||
|
||||
FABRIX_ENV = {k: _env(k) for k in _FABRIX_KEYS}
|
||||
# Langfuse(LLM 관측) — 자체 호스팅 주소 + 프로젝트 키. HOST 비우면 전송 안 함
|
||||
LANGFUSE = {
|
||||
"host": _env("LANGFUSE_HOST"),
|
||||
"public_key": _env("LANGFUSE_PUBLIC_KEY"),
|
||||
"secret_key": _env("LANGFUSE_SECRET_KEY"),
|
||||
}
|
||||
# 세션 컨텍스트 하드 한도 — usage.limit 로 프론트 게이지에 감
|
||||
CONTEXT_LIMIT_TOKENS = int(_env("CONTEXT_LIMIT_TOKENS", "128000") or "128000")
|
||||
|
||||
|
||||
@@ -19,6 +19,7 @@ grep -q '^BACKEND_PORT=' .env || echo 'BACKEND_PORT=8080' >> .env
|
||||
grep -q '^OPENCODE_VISION_MODEL=' .env || echo 'OPENCODE_VISION_MODEL=gateway/605' >> .env
|
||||
# 고객사 FabriX 는 Bearer 접두 필수(9/18 실측). 첫 시도부터 맞게 — 날것으로 보내면 401 뒤 재시도에 기댐
|
||||
grep -q '^AAF_FABRIX_TOKEN_PREFIX=' .env || echo 'AAF_FABRIX_TOKEN_PREFIX=bearer' >> .env # OpenCode 가 붙을 Django 포트. 없으면 8001 로 렌더돼 'Unable to connect'
|
||||
for k in LANGFUSE_HOST LANGFUSE_PUBLIC_KEY LANGFUSE_SECRET_KEY; do grep -q "^$k=" .env || echo "$k=" >> .env; done # 비우면 관측 안 함
|
||||
sed -i 's/^AAF_FABRIX_MODEL_ID=.*/AAF_FABRIX_MODEL_ID=581/; s/^AAF_FABRIX_MODELS=.*/AAF_FABRIX_MODELS=581:GaussO Think,339:GaussO Flash,605:Gemma4/' .env
|
||||
.venv/bin/python manage.py migrate --noinput
|
||||
[ -f seed/snippets.db ] && .venv/bin/python manage.py import_snippets seed/snippets.db
|
||||
|
||||
@@ -0,0 +1,24 @@
|
||||
# Langfuse 자체 호스팅 (.env — docker-compose.yml 이 읽음). 전부 CHANGEME 는 바꿔서 쓸 것.
|
||||
# 값 만드는 법: openssl rand -hex 32 (SALT·NEXTAUTH_SECRET·ENCRYPTION_KEY 각각 다르게)
|
||||
|
||||
# 웹 UI 가 열리는 주소 — 고객사는 -12 남는 포트(예: 8915→3000)로. 로컬은 http://localhost:3000
|
||||
NEXTAUTH_URL=http://localhost:3000
|
||||
NEXTAUTH_SECRET=CHANGEME
|
||||
SALT=CHANGEME
|
||||
ENCRYPTION_KEY=CHANGEME_HEX64
|
||||
|
||||
# 컨테이너 안 서비스(postgres/clickhouse/redis/minio) 비번은 compose 기본값 그대로 둠 —
|
||||
# 밖으로 안 열리는 내부 네트워크라 위험 없고, 바꾸면 DATABASE_URL·LANGFUSE_S3_* 도 같이 맞춰야 해서 실수 잦음.
|
||||
|
||||
# 헤드리스 초기화 — 첫 기동 때 조직/프로젝트/키/관리자 자동 생성. 이 pk/sk 를 백엔드 .env 의 LANGFUSE_* 에 그대로.
|
||||
LANGFUSE_INIT_ORG_ID=codeassist
|
||||
LANGFUSE_INIT_ORG_NAME=CodeAssist
|
||||
LANGFUSE_INIT_PROJECT_ID=codeassist
|
||||
LANGFUSE_INIT_PROJECT_NAME=CodeAssist
|
||||
LANGFUSE_INIT_PROJECT_PUBLIC_KEY=pk-lf-CHANGEME
|
||||
LANGFUSE_INIT_PROJECT_SECRET_KEY=sk-lf-CHANGEME
|
||||
LANGFUSE_INIT_USER_EMAIL=admin@example.com
|
||||
LANGFUSE_INIT_USER_NAME=admin
|
||||
LANGFUSE_INIT_USER_PASSWORD=CHANGEME
|
||||
|
||||
TELEMETRY_ENABLED=false
|
||||
@@ -0,0 +1,41 @@
|
||||
# Langfuse 자체 호스팅 — -12 에 올리기
|
||||
|
||||
LLM 호출 관측(누가·언제·어떤 모델·프롬프트·답·토큰·시간). 백엔드가 `LANGFUSE_HOST` 있으면 자동으로 쏨(`apps/gateway/langfuse.py`).
|
||||
공식 `docker-compose.yml` 그대로(2026-09-22 main). 컨테이너 6개, 밖으로 여는 건 웹 3000 하나.
|
||||
|
||||
## 1. 이미지 반입 (인터넷 되는 PC)
|
||||
|
||||
```bash
|
||||
cd 5_django_backend/deploy/langfuse
|
||||
docker compose pull
|
||||
docker save $(grep -oE 'image: \S+' docker-compose.yml | cut -d' ' -f2 | sed 's/\${POSTGRES_VERSION:-17}/17/') | gzip > langfuse-images.tar.gz
|
||||
ls -la langfuse-images.tar.gz # 3~4GB. 메일 안 됨 — 인프라팀 반입 통로로
|
||||
```
|
||||
|
||||
## 2. -12 에서 띄우기
|
||||
|
||||
```bash
|
||||
docker load < langfuse-images.tar.gz
|
||||
cd /www/abap-ito/code-assistant/5_django_backend/deploy/langfuse
|
||||
cp .env.example .env && vi .env # CHANGEME 전부 교체. NEXTAUTH_URL 은 사용자가 여는 주소
|
||||
docker compose up -d
|
||||
docker compose ps # 6개 healthy 까지 1~2분
|
||||
curl -s http://127.0.0.1:3000/api/public/health
|
||||
```
|
||||
|
||||
## 3. 백엔드 연결
|
||||
|
||||
`5_django_backend/.env` 에 (`.env` 의 INIT 키와 같은 값):
|
||||
```
|
||||
LANGFUSE_HOST=http://127.0.0.1:3000
|
||||
LANGFUSE_PUBLIC_KEY=pk-lf-…
|
||||
LANGFUSE_SECRET_KEY=sk-lf-…
|
||||
```
|
||||
`bash deploy.sh` 로 재기동 → 채팅 하나 보내고 Langfuse 웹(Traces)에 `chat`·`fabrix` trace 뜨면 끝.
|
||||
|
||||
## 운영
|
||||
|
||||
- 웹 UI 접속은 -12 의 3000 → 인프라팀이 외부 포트(8915 등) 매핑해줘야 고객사 PC 에서 열림
|
||||
- 중지 `docker compose down` (데이터 볼륨 남음) / 완전 삭제 `docker compose down -v`
|
||||
- 로그 `docker compose logs -f langfuse-web`
|
||||
- 백엔드는 Langfuse 죽어 있어도 멀쩡(전송 실패는 warning 로그만)
|
||||
@@ -0,0 +1,225 @@
|
||||
# Make sure to update the credential placeholders with your own secrets.
|
||||
# We mark them with # CHANGEME in the file below.
|
||||
# In addition, we recommend to restrict inbound traffic on the host to langfuse-web (port 3000) and minio (port 9090) only.
|
||||
# All other components are bound to localhost (127.0.0.1) to only accept connections from the local machine.
|
||||
# External connections from other machines will not be able to reach these services directly.
|
||||
services:
|
||||
langfuse-worker:
|
||||
image: docker.langfuse.com/langfuse/langfuse-worker:4
|
||||
restart: always
|
||||
depends_on: &langfuse-depends-on
|
||||
postgres:
|
||||
condition: service_healthy
|
||||
minio:
|
||||
condition: service_healthy
|
||||
redis:
|
||||
condition: service_healthy
|
||||
clickhouse:
|
||||
condition: service_healthy
|
||||
ports:
|
||||
- 127.0.0.1:3030:3030
|
||||
environment: &langfuse-worker-env
|
||||
NEXTAUTH_URL: ${NEXTAUTH_URL:-http://localhost:3000}
|
||||
DATABASE_URL: ${DATABASE_URL:-postgresql://postgres:postgres@postgres:5432/postgres} # CHANGEME
|
||||
SALT: ${SALT:-mysalt} # CHANGEME
|
||||
ENCRYPTION_KEY: ${ENCRYPTION_KEY:-0000000000000000000000000000000000000000000000000000000000000000} # CHANGEME: generate via `openssl rand -hex 32`
|
||||
TELEMETRY_ENABLED: ${TELEMETRY_ENABLED:-true}
|
||||
LANGFUSE_ENABLE_EXPERIMENTAL_FEATURES: ${LANGFUSE_ENABLE_EXPERIMENTAL_FEATURES:-false}
|
||||
# Lets the worker reach the web service internally for in-app agent MCP
|
||||
# calls without redirecting NEXTAUTH_URL, which also builds links for
|
||||
# users. Pair it with LANGFUSE_MCP_ALLOWED_HOSTS on langfuse-web.
|
||||
LANGFUSE_MCP_BASE_URL: ${LANGFUSE_MCP_BASE_URL:-}
|
||||
LANGFUSE_LLM_CONNECTION_WHITELISTED_HOST: ${LANGFUSE_LLM_CONNECTION_WHITELISTED_HOST:-}
|
||||
LANGFUSE_LLM_CONNECTION_WHITELISTED_IPS: ${LANGFUSE_LLM_CONNECTION_WHITELISTED_IPS:-}
|
||||
LANGFUSE_LLM_CONNECTION_WHITELISTED_IP_SEGMENTS: ${LANGFUSE_LLM_CONNECTION_WHITELISTED_IP_SEGMENTS:-}
|
||||
CLICKHOUSE_MIGRATION_URL: ${CLICKHOUSE_MIGRATION_URL:-clickhouse://clickhouse:9000}
|
||||
CLICKHOUSE_URL: ${CLICKHOUSE_URL:-http://clickhouse:8123}
|
||||
CLICKHOUSE_USER: ${CLICKHOUSE_USER:-clickhouse}
|
||||
CLICKHOUSE_PASSWORD: ${CLICKHOUSE_PASSWORD:-clickhouse} # CHANGEME
|
||||
CLICKHOUSE_CLUSTER_ENABLED: ${CLICKHOUSE_CLUSTER_ENABLED:-false}
|
||||
CLICKHOUSE_CLUSTER_NAME: ${CLICKHOUSE_CLUSTER_NAME:-default}
|
||||
LANGFUSE_JSON_BAD_UNICODE_ESCAPE: ${LANGFUSE_JSON_BAD_UNICODE_ESCAPE:-}
|
||||
LANGFUSE_USE_AZURE_BLOB: ${LANGFUSE_USE_AZURE_BLOB:-false}
|
||||
LANGFUSE_USE_OCI_NATIVE_OBJECT_STORAGE: ${LANGFUSE_USE_OCI_NATIVE_OBJECT_STORAGE:-false}
|
||||
LANGFUSE_OCI_AUTH_TYPE: ${LANGFUSE_OCI_AUTH_TYPE:-workload_identity}
|
||||
LANGFUSE_S3_EVENT_UPLOAD_BUCKET: ${LANGFUSE_S3_EVENT_UPLOAD_BUCKET:-langfuse}
|
||||
LANGFUSE_S3_EVENT_UPLOAD_REGION: ${LANGFUSE_S3_EVENT_UPLOAD_REGION:-auto}
|
||||
LANGFUSE_S3_EVENT_UPLOAD_ACCESS_KEY_ID: ${LANGFUSE_S3_EVENT_UPLOAD_ACCESS_KEY_ID:-minio}
|
||||
LANGFUSE_S3_EVENT_UPLOAD_SECRET_ACCESS_KEY: ${LANGFUSE_S3_EVENT_UPLOAD_SECRET_ACCESS_KEY:-miniosecret} # CHANGEME
|
||||
LANGFUSE_S3_EVENT_UPLOAD_ENDPOINT: ${LANGFUSE_S3_EVENT_UPLOAD_ENDPOINT:-http://minio:9000}
|
||||
LANGFUSE_S3_EVENT_UPLOAD_FORCE_PATH_STYLE: ${LANGFUSE_S3_EVENT_UPLOAD_FORCE_PATH_STYLE:-true}
|
||||
LANGFUSE_S3_EVENT_UPLOAD_PREFIX: ${LANGFUSE_S3_EVENT_UPLOAD_PREFIX:-events/}
|
||||
LANGFUSE_S3_MEDIA_UPLOAD_BUCKET: ${LANGFUSE_S3_MEDIA_UPLOAD_BUCKET:-langfuse}
|
||||
LANGFUSE_S3_MEDIA_UPLOAD_REGION: ${LANGFUSE_S3_MEDIA_UPLOAD_REGION:-auto}
|
||||
LANGFUSE_S3_MEDIA_UPLOAD_ACCESS_KEY_ID: ${LANGFUSE_S3_MEDIA_UPLOAD_ACCESS_KEY_ID:-minio}
|
||||
LANGFUSE_S3_MEDIA_UPLOAD_SECRET_ACCESS_KEY: ${LANGFUSE_S3_MEDIA_UPLOAD_SECRET_ACCESS_KEY:-miniosecret} # CHANGEME
|
||||
LANGFUSE_S3_MEDIA_UPLOAD_ENDPOINT: ${LANGFUSE_S3_MEDIA_UPLOAD_ENDPOINT:-http://minio:9000}
|
||||
LANGFUSE_S3_MEDIA_UPLOAD_FORCE_PATH_STYLE: ${LANGFUSE_S3_MEDIA_UPLOAD_FORCE_PATH_STYLE:-true}
|
||||
LANGFUSE_S3_MEDIA_UPLOAD_PREFIX: ${LANGFUSE_S3_MEDIA_UPLOAD_PREFIX:-media/}
|
||||
LANGFUSE_OBSERVATION_FIELD_OVERFLOW_ENABLED: ${LANGFUSE_OBSERVATION_FIELD_OVERFLOW_ENABLED:-false}
|
||||
LANGFUSE_OBSERVATION_FIELD_SIZE_LIMIT_BYTES: ${LANGFUSE_OBSERVATION_FIELD_SIZE_LIMIT_BYTES:-2097152}
|
||||
LANGFUSE_S3_BATCH_EXPORT_ENABLED: ${LANGFUSE_S3_BATCH_EXPORT_ENABLED:-false}
|
||||
LANGFUSE_S3_BATCH_EXPORT_BUCKET: ${LANGFUSE_S3_BATCH_EXPORT_BUCKET:-langfuse}
|
||||
LANGFUSE_S3_BATCH_EXPORT_PREFIX: ${LANGFUSE_S3_BATCH_EXPORT_PREFIX:-exports/}
|
||||
LANGFUSE_S3_BATCH_EXPORT_REGION: ${LANGFUSE_S3_BATCH_EXPORT_REGION:-auto}
|
||||
LANGFUSE_S3_BATCH_EXPORT_ENDPOINT: ${LANGFUSE_S3_BATCH_EXPORT_ENDPOINT:-http://minio:9000}
|
||||
LANGFUSE_S3_BATCH_EXPORT_EXTERNAL_ENDPOINT: ${LANGFUSE_S3_BATCH_EXPORT_EXTERNAL_ENDPOINT:-http://localhost:9090}
|
||||
LANGFUSE_S3_BATCH_EXPORT_ACCESS_KEY_ID: ${LANGFUSE_S3_BATCH_EXPORT_ACCESS_KEY_ID:-minio}
|
||||
LANGFUSE_S3_BATCH_EXPORT_SECRET_ACCESS_KEY: ${LANGFUSE_S3_BATCH_EXPORT_SECRET_ACCESS_KEY:-miniosecret} # CHANGEME
|
||||
LANGFUSE_S3_BATCH_EXPORT_FORCE_PATH_STYLE: ${LANGFUSE_S3_BATCH_EXPORT_FORCE_PATH_STYLE:-true}
|
||||
LANGFUSE_INGESTION_QUEUE_DELAY_MS: ${LANGFUSE_INGESTION_QUEUE_DELAY_MS:-}
|
||||
LANGFUSE_INGESTION_CLICKHOUSE_WRITE_INTERVAL_MS: ${LANGFUSE_INGESTION_CLICKHOUSE_WRITE_INTERVAL_MS:-}
|
||||
REDIS_HOST: ${REDIS_HOST:-redis}
|
||||
REDIS_PORT: ${REDIS_PORT:-6379}
|
||||
REDIS_AUTH: ${REDIS_AUTH:-myredissecret} # CHANGEME
|
||||
LANGFUSE_BULLMQ_SKIP_REDIS_VERSION_CHECK: ${LANGFUSE_BULLMQ_SKIP_REDIS_VERSION_CHECK:-false}
|
||||
REDIS_TLS_ENABLED: ${REDIS_TLS_ENABLED:-false}
|
||||
REDIS_TLS_CA: ${REDIS_TLS_CA:-/certs/ca.crt}
|
||||
REDIS_TLS_CERT: ${REDIS_TLS_CERT:-/certs/redis.crt}
|
||||
REDIS_TLS_KEY: ${REDIS_TLS_KEY:-/certs/redis.key}
|
||||
EMAIL_FROM_ADDRESS: ${EMAIL_FROM_ADDRESS:-}
|
||||
SMTP_CONNECTION_URL: ${SMTP_CONNECTION_URL:-}
|
||||
# Shared AI features configuration. These apply to AI features that run
|
||||
# in both web and worker.
|
||||
LANGFUSE_IN_APP_AGENT_ENABLED: ${LANGFUSE_IN_APP_AGENT_ENABLED:-}
|
||||
LANGFUSE_EVALUATOR_MEDIA_TRANSPORT: ${LANGFUSE_EVALUATOR_MEDIA_TRANSPORT:-}
|
||||
LANGFUSE_EVALUATOR_MEDIA_INLINE_MAX_BYTES: ${LANGFUSE_EVALUATOR_MEDIA_INLINE_MAX_BYTES:-}
|
||||
LANGFUSE_AI_PROVIDER: ${LANGFUSE_AI_PROVIDER:-}
|
||||
LANGFUSE_AI_MODEL: ${LANGFUSE_AI_MODEL:-}
|
||||
LANGFUSE_AI_SMALL_MODEL: ${LANGFUSE_AI_SMALL_MODEL:-}
|
||||
LANGFUSE_AI_API_KEY: ${LANGFUSE_AI_API_KEY:-}
|
||||
LANGFUSE_AI_BASE_URL: ${LANGFUSE_AI_BASE_URL:-}
|
||||
LANGFUSE_AI_USE_RESPONSES_API: ${LANGFUSE_AI_USE_RESPONSES_API:-}
|
||||
LANGFUSE_AI_EXTRA_HEADERS: ${LANGFUSE_AI_EXTRA_HEADERS:-}
|
||||
LANGFUSE_AI_AWS_BEDROCK_REGION: ${LANGFUSE_AI_AWS_BEDROCK_REGION:-}
|
||||
LANGFUSE_AI_VERTEX_LOCATION: ${LANGFUSE_AI_VERTEX_LOCATION:-}
|
||||
AWS_ACCESS_KEY_ID: ${AWS_ACCESS_KEY_ID:-}
|
||||
AWS_SECRET_ACCESS_KEY: ${AWS_SECRET_ACCESS_KEY:-}
|
||||
AWS_SESSION_TOKEN: ${AWS_SESSION_TOKEN:-}
|
||||
AWS_PROFILE: ${AWS_PROFILE:-}
|
||||
# In-app agent worker settings (ignored by the web service).
|
||||
LANGFUSE_IN_APP_AGENT_RUN_QUEUE_PROCESSING_CONCURRENCY: ${LANGFUSE_IN_APP_AGENT_RUN_QUEUE_PROCESSING_CONCURRENCY:-}
|
||||
LANGFUSE_IN_APP_AGENT_AWS_PROFILE: ${LANGFUSE_IN_APP_AGENT_AWS_PROFILE:-}
|
||||
LANGFUSE_IN_APP_AGENT_SANDBOX_PROVIDER: ${LANGFUSE_IN_APP_AGENT_SANDBOX_PROVIDER:-}
|
||||
LANGFUSE_IN_APP_AGENT_SANDBOX_AWS_LAMBDA_MICROVM_IMAGE_IDENTIFIER: ${LANGFUSE_IN_APP_AGENT_SANDBOX_AWS_LAMBDA_MICROVM_IMAGE_IDENTIFIER:-}
|
||||
LANGFUSE_IN_APP_AGENT_SANDBOX_AWS_LAMBDA_MICROVM_EXECUTION_ROLE_ARN: ${LANGFUSE_IN_APP_AGENT_SANDBOX_AWS_LAMBDA_MICROVM_EXECUTION_ROLE_ARN:-}
|
||||
LANGFUSE_IN_APP_AGENT_SANDBOX_AWS_LAMBDA_MICROVM_EGRESS_NETWORK_CONNECTOR_ARN: ${LANGFUSE_IN_APP_AGENT_SANDBOX_AWS_LAMBDA_MICROVM_EGRESS_NETWORK_CONNECTOR_ARN:-}
|
||||
LANGFUSE_IN_APP_AGENT_SANDBOX_AWS_LAMBDA_MICROVM_REGION: ${LANGFUSE_IN_APP_AGENT_SANDBOX_AWS_LAMBDA_MICROVM_REGION:-}
|
||||
|
||||
langfuse-web:
|
||||
image: docker.langfuse.com/langfuse/langfuse:4
|
||||
restart: always
|
||||
depends_on: *langfuse-depends-on
|
||||
ports:
|
||||
- 3000:3000
|
||||
environment:
|
||||
<<: *langfuse-worker-env
|
||||
NEXTAUTH_SECRET: ${NEXTAUTH_SECRET:-mysecret} # CHANGEME
|
||||
LANGFUSE_S3_MEDIA_UPLOAD_ENDPOINT: ${LANGFUSE_S3_MEDIA_UPLOAD_ENDPOINT:-http://localhost:9090}
|
||||
LANGFUSE_S3_MEDIA_UPLOAD_INTERNAL_ENDPOINT: ${LANGFUSE_S3_MEDIA_UPLOAD_INTERNAL_ENDPOINT:-http://minio:9000}
|
||||
LANGFUSE_INIT_ORG_ID: ${LANGFUSE_INIT_ORG_ID:-}
|
||||
LANGFUSE_INIT_ORG_NAME: ${LANGFUSE_INIT_ORG_NAME:-}
|
||||
LANGFUSE_INIT_PROJECT_ID: ${LANGFUSE_INIT_PROJECT_ID:-}
|
||||
LANGFUSE_INIT_PROJECT_NAME: ${LANGFUSE_INIT_PROJECT_NAME:-}
|
||||
LANGFUSE_INIT_PROJECT_PUBLIC_KEY: ${LANGFUSE_INIT_PROJECT_PUBLIC_KEY:-}
|
||||
LANGFUSE_INIT_PROJECT_SECRET_KEY: ${LANGFUSE_INIT_PROJECT_SECRET_KEY:-}
|
||||
LANGFUSE_INIT_USER_EMAIL: ${LANGFUSE_INIT_USER_EMAIL:-}
|
||||
LANGFUSE_INIT_USER_NAME: ${LANGFUSE_INIT_USER_NAME:-}
|
||||
LANGFUSE_INIT_USER_PASSWORD: ${LANGFUSE_INIT_USER_PASSWORD:-}
|
||||
LANGFUSE_MARKDOWN_RENDER_CHARACTER_LIMIT: ${LANGFUSE_MARKDOWN_RENDER_CHARACTER_LIMIT:-}
|
||||
# Hostnames allowed to reach the MCP endpoint, beyond NEXTAUTH_URL's own
|
||||
# host. Required when the worker uses LANGFUSE_MCP_BASE_URL.
|
||||
LANGFUSE_MCP_ALLOWED_HOSTS: ${LANGFUSE_MCP_ALLOWED_HOSTS:-}
|
||||
# In-app agent settings.
|
||||
LANGFUSE_IN_APP_AGENT_MAX_ACTIVE_RUNS_PER_USER: ${LANGFUSE_IN_APP_AGENT_MAX_ACTIVE_RUNS_PER_USER:-}
|
||||
LANGFUSE_IN_APP_AGENT_MAX_ACTIVE_RUNS_PER_ORG: ${LANGFUSE_IN_APP_AGENT_MAX_ACTIVE_RUNS_PER_ORG:-}
|
||||
|
||||
clickhouse:
|
||||
image: docker.io/clickhouse/clickhouse-server:25.12
|
||||
restart: always
|
||||
user: "101:101"
|
||||
environment:
|
||||
CLICKHOUSE_DB: default
|
||||
CLICKHOUSE_USER: ${CLICKHOUSE_USER:-clickhouse}
|
||||
CLICKHOUSE_PASSWORD: ${CLICKHOUSE_PASSWORD:-clickhouse} # CHANGEME
|
||||
volumes:
|
||||
- langfuse_clickhouse_data:/var/lib/clickhouse
|
||||
- langfuse_clickhouse_logs:/var/log/clickhouse-server
|
||||
ports:
|
||||
- 127.0.0.1:8123:8123
|
||||
- 127.0.0.1:9000:9000
|
||||
healthcheck:
|
||||
test: wget --no-verbose --tries=1 --spider http://localhost:8123/ping || exit 1
|
||||
interval: 5s
|
||||
timeout: 5s
|
||||
retries: 10
|
||||
start_period: 1s
|
||||
|
||||
minio:
|
||||
image: cgr.dev/chainguard/minio
|
||||
restart: always
|
||||
entrypoint: sh
|
||||
# create the 'langfuse' bucket before starting the service
|
||||
command: -c 'mkdir -p /data/langfuse && minio server --address ":9000" --console-address ":9001" /data'
|
||||
environment:
|
||||
MINIO_ROOT_USER: ${MINIO_ROOT_USER:-minio}
|
||||
MINIO_ROOT_PASSWORD: ${MINIO_ROOT_PASSWORD:-miniosecret} # CHANGEME
|
||||
ports:
|
||||
- 9090:9000
|
||||
- 127.0.0.1:9091:9001
|
||||
volumes:
|
||||
- langfuse_minio_data:/data
|
||||
healthcheck:
|
||||
test: ["CMD", "mc", "ready", "local"]
|
||||
interval: 1s
|
||||
timeout: 5s
|
||||
retries: 5
|
||||
start_period: 1s
|
||||
|
||||
redis:
|
||||
image: docker.io/redis:7
|
||||
restart: always
|
||||
# CHANGEME: row below to secure redis password
|
||||
command: >
|
||||
--requirepass ${REDIS_AUTH:-myredissecret}
|
||||
--maxmemory-policy noeviction
|
||||
ports:
|
||||
- 127.0.0.1:6379:6379
|
||||
volumes:
|
||||
- langfuse_redis_data:/data
|
||||
healthcheck:
|
||||
test: ["CMD", "redis-cli", "ping"]
|
||||
interval: 3s
|
||||
timeout: 10s
|
||||
retries: 10
|
||||
|
||||
postgres:
|
||||
image: docker.io/postgres:${POSTGRES_VERSION:-17}
|
||||
restart: always
|
||||
healthcheck:
|
||||
test: ["CMD-SHELL", "pg_isready -U postgres"]
|
||||
interval: 3s
|
||||
timeout: 3s
|
||||
retries: 10
|
||||
environment:
|
||||
POSTGRES_USER: ${POSTGRES_USER:-postgres}
|
||||
POSTGRES_PASSWORD: ${POSTGRES_PASSWORD:-postgres} # CHANGEME
|
||||
POSTGRES_DB: ${POSTGRES_DB:-postgres}
|
||||
TZ: UTC
|
||||
PGTZ: UTC
|
||||
ports:
|
||||
- 127.0.0.1:5432:5432
|
||||
volumes:
|
||||
- langfuse_postgres_data:/var/lib/postgresql/data
|
||||
|
||||
volumes:
|
||||
langfuse_postgres_data:
|
||||
driver: local
|
||||
langfuse_clickhouse_data:
|
||||
driver: local
|
||||
langfuse_clickhouse_logs:
|
||||
driver: local
|
||||
langfuse_minio_data:
|
||||
driver: local
|
||||
langfuse_redis_data:
|
||||
driver: local
|
||||
@@ -4,6 +4,7 @@
|
||||
|
||||
| 파일 | 뭐냐 | 출처 |
|
||||
|---|---|---|
|
||||
| `langfuse.md` | 자체 호스팅 Langfuse ingestion API(배치 이벤트 모양·인증·207) + docker-compose env·헤드리스 초기화 키 | openapi.yml, langfuse.com/self-hosting (2026-09-22) |
|
||||
| `opencode-sdk-1.18.6-types.d.ts` | OpenCode 서버(`opencode serve`) HTTP API 전체 타입 — 세션·메시지·파트·이벤트·엔드포인트 url. `url: "/session…"` 로 grep 하면 엔드포인트, `export type Event…` 로 grep 하면 SSE 이벤트 | `@opencode-ai/sdk@1.18.6` `dist/gen/types.gen.d.ts` (ABAP_OPENCODE `.opencode/node_modules`) |
|
||||
|
||||
갱신: OpenCode 버전 올리면 같은 경로에서 다시 복사하고 파일명 버전 바꿈.
|
||||
|
||||
@@ -0,0 +1,39 @@
|
||||
# Langfuse — 자체 호스팅 ingestion API (2026-09-22 정리)
|
||||
|
||||
출처: `https://cloud.langfuse.com/generated/api/openapi.yml`, `langfuse.com/self-hosting`. SDK 안 쓰고 HTTP 로 직접 쏨(오프라인 wheels 반입 줄이려고).
|
||||
|
||||
## POST /api/public/ingestion
|
||||
|
||||
- 인증: Basic auth — user = `pk-lf-…`(public key), password = `sk-lf-…`(secret key)
|
||||
- 응답: **207** (배치 부분 성공). 각 이벤트 결과가 `successes[]`/`errors[]` 로 옴
|
||||
- Langfuse Cloud 에선 2026-11-16 폐기 예정이지만 **self-host 는 계속 지원**. OTel 엔드포인트(`/api/public/otel/v1/traces`)는 v3+ 만.
|
||||
|
||||
```json
|
||||
{
|
||||
"batch": [
|
||||
{"id": "evt-1", "timestamp": "2026-09-22T01:00:00Z", "type": "trace-create",
|
||||
"body": {"id": "trace-1", "name": "chat", "userId": "u@x.com", "sessionId": "ses_…", "input": "…", "output": "…", "metadata": {}, "tags": []}},
|
||||
{"id": "evt-2", "timestamp": "2026-09-22T01:00:05Z", "type": "generation-create",
|
||||
"body": {"id": "gen-1", "traceId": "trace-1", "name": "fabrix", "model": "581",
|
||||
"startTime": "…", "endTime": "…", "input": [...messages], "output": "…",
|
||||
"usage": {"input": 120, "output": 30, "total": 150}, "level": "DEFAULT", "statusMessage": "", "metadata": {}}}
|
||||
]
|
||||
}
|
||||
```
|
||||
|
||||
- envelope: `id`(중복 제거용, 유일), `timestamp`(ISO 8601), `type`, `body`
|
||||
- 이벤트 타입: `trace-create`, `generation-create`, `span-create`, `event-create`, `score-create`, `*-update`
|
||||
- TraceBody: id·name·userId·sessionId·input·output·metadata·tags·version·release·environment·public
|
||||
- CreateGenerationBody: id·traceId·name·startTime·endTime·completionStartTime·model·modelParameters·input·output·usage{input,output,total,unit,inputCost,outputCost,totalCost}·usageDetails·costDetails·level(DEBUG/DEFAULT/WARNING/ERROR)·statusMessage·metadata·parentObservationId
|
||||
- 같은 traceId 로 trace-create 를 여러 번 보내도 됨(upsert). generation 먼저 와도 trace 가 나중에 붙음.
|
||||
|
||||
## 자체 호스팅 (v4 docker-compose, 2026-09 기준)
|
||||
|
||||
이미지 6개: `langfuse/langfuse:4`, `langfuse/langfuse-worker:4`, `clickhouse/clickhouse-server:25.12`, `cgr.dev/chainguard/minio`, `redis:7`, `postgres:17`.
|
||||
web 포트 3000. 필수 env: `DATABASE_URL`, `NEXTAUTH_URL`, `NEXTAUTH_SECRET`, `SALT`, `ENCRYPTION_KEY`(hex 64자), CLICKHOUSE_*, REDIS_*, LANGFUSE_S3_*.
|
||||
|
||||
헤드리스 초기화(첫 기동 때 조직·프로젝트·키·관리자 자동 생성, UI 클릭 없이):
|
||||
`LANGFUSE_INIT_ORG_ID`, `LANGFUSE_INIT_ORG_NAME`, `LANGFUSE_INIT_PROJECT_ID`, `LANGFUSE_INIT_PROJECT_NAME`,
|
||||
`LANGFUSE_INIT_PROJECT_PUBLIC_KEY`(pk-lf-…), `LANGFUSE_INIT_PROJECT_SECRET_KEY`(sk-lf-…),
|
||||
`LANGFUSE_INIT_USER_EMAIL`, `LANGFUSE_INIT_USER_NAME`, `LANGFUSE_INIT_USER_PASSWORD`.
|
||||
→ 우리 `deploy/langfuse/.env` 가 이걸 채우고, 같은 pk/sk 를 백엔드 `.env` 의 LANGFUSE_* 에 넣음.
|
||||
@@ -126,6 +126,22 @@ async def test_401_retries_other_auth_shape_and_remembers(upstream):
|
||||
assert len(calls) == n + 1 and calls[-1].headers["x-openapi-token"] == "Bearer tok"
|
||||
|
||||
|
||||
@override_settings(FABRIX_ENV=FULL_ENV, LANGFUSE={"host": "http://lf.test", "public_key": "pk", "secret_key": "sk"})
|
||||
async def test_stream_reports_generation_to_langfuse(upstream, monkeypatch):
|
||||
import asyncio
|
||||
|
||||
from apps.gateway import langfuse
|
||||
|
||||
got = []
|
||||
monkeypatch.setattr(langfuse, "TRANSPORT", httpx.MockTransport(lambda r: (got.append(json.loads(r.content)), httpx.Response(207, json={}))[1]))
|
||||
upstream(lambda r: _sse(b'data: {"choices":[{"delta":{"content":"hi"}}],"usage":{"prompt_tokens":2,"completion_tokens":1}}\n\ndata: [DONE]\n\n'))
|
||||
resp = await _post({"model": "581", "stream": True, "messages": [{"role": "user", "content": "q"}]})
|
||||
b"".join([c async for c in resp.streaming_content])
|
||||
await asyncio.gather(*list(langfuse._pending))
|
||||
gen = got[0]["batch"][1]["body"]
|
||||
assert gen["model"] == "581" and gen["output"] == "hi" and gen["usage"]["total"] == 3 and gen["input"][0]["content"] == "q"
|
||||
|
||||
|
||||
@override_settings(FABRIX_ENV=FULL_ENV)
|
||||
async def test_flaky_401_on_right_auth_is_retried_once(upstream, monkeypatch):
|
||||
"""FabriX 가 맞는 조합에도 가끔 401 — 4개 다 돈 뒤 맨 앞 조합 한 번 더."""
|
||||
|
||||
@@ -0,0 +1,87 @@
|
||||
"""Langfuse 전송 — 이벤트 모양·인증·게이트웨이/턴 마무리 훅. 서버는 MockTransport."""
|
||||
|
||||
import asyncio
|
||||
import base64
|
||||
import json
|
||||
|
||||
import httpx
|
||||
import pytest
|
||||
from django.test import override_settings
|
||||
|
||||
from apps.gateway import langfuse
|
||||
|
||||
LF = {"host": "http://lf.test", "public_key": "pk-lf-a", "secret_key": "sk-lf-b"}
|
||||
|
||||
|
||||
@pytest.fixture
|
||||
def lf_server(monkeypatch):
|
||||
got: list[dict] = []
|
||||
|
||||
def handler(req: httpx.Request) -> httpx.Response:
|
||||
got.append({"url": str(req.url), "auth": req.headers.get("authorization"), "body": json.loads(req.content)})
|
||||
return httpx.Response(207, json={"successes": [], "errors": []})
|
||||
|
||||
monkeypatch.setattr(langfuse, "TRANSPORT", httpx.MockTransport(handler))
|
||||
return got
|
||||
|
||||
|
||||
@override_settings(LANGFUSE=LF)
|
||||
async def test_send_batch_shape_and_basic_auth(lf_server):
|
||||
ok = await langfuse.send([
|
||||
langfuse.trace("t1", "chat", userId="u@x.com", sessionId="s1"),
|
||||
langfuse.generation("t1", "gen", model="581", usage=langfuse.usage_of(10, 5)),
|
||||
])
|
||||
assert ok and len(lf_server) == 1
|
||||
req = lf_server[0]
|
||||
assert req["url"] == "http://lf.test/api/public/ingestion"
|
||||
assert req["auth"] == "Basic " + base64.b64encode(b"pk-lf-a:sk-lf-b").decode()
|
||||
batch = req["body"]["batch"]
|
||||
assert [e["type"] for e in batch] == ["trace-create", "generation-create"]
|
||||
assert batch[0]["body"] == {"id": "t1", "name": "chat", "userId": "u@x.com", "sessionId": "s1"}
|
||||
assert batch[1]["body"]["traceId"] == "t1" and batch[1]["body"]["usage"] == {"input": 10, "output": 5, "total": 15}
|
||||
assert batch[0]["id"] != batch[1]["id"] and batch[0]["timestamp"].endswith("Z")
|
||||
|
||||
|
||||
@override_settings(LANGFUSE={"host": "", "public_key": "", "secret_key": ""})
|
||||
async def test_disabled_is_noop(lf_server):
|
||||
assert not langfuse.enabled()
|
||||
assert await langfuse.send([langfuse.trace("t", "x")]) is False
|
||||
langfuse.send_later([langfuse.trace("t", "x")])
|
||||
assert lf_server == []
|
||||
|
||||
|
||||
def test_collect_parses_sse_text_and_usage():
|
||||
from apps.gateway.views import _Collect
|
||||
|
||||
c = _Collect()
|
||||
c.feed(b'data: {"choices":[{"delta":{"content":"SEL"}}]}\n\ndata: {"choices":[{"delta":{"co')
|
||||
c.feed(b'ntent":"ECT"}}],"usage":{"prompt_tokens":7,"completion_tokens":2}}\n\ndata: [DONE]\n\n')
|
||||
assert "".join(c.text) == "SELECT" and c.usage == {"prompt_tokens": 7, "completion_tokens": 2}
|
||||
|
||||
|
||||
@pytest.mark.django_db(transaction=True)
|
||||
async def test_finalize_sends_user_trace(lf_server, django_user_model, monkeypatch):
|
||||
from apps.chat import stream as stream_mod
|
||||
from apps.chat.models import ChatSession
|
||||
from apps.chat.stream import TurnState, _finalize
|
||||
from common.opencode_service import opencode_service
|
||||
|
||||
async def _msgs(_sid):
|
||||
return [{"info": {"role": "assistant", "tokens": {"input": 3, "output": 4, "reasoning": 1}, "time": {"created": 1000, "completed": 1500}}, "parts": [{"type": "text", "text": "답"}]}]
|
||||
|
||||
async def _sess(_sid):
|
||||
return {"title": "MARA 조회"}
|
||||
|
||||
monkeypatch.setattr(opencode_service, "list_messages_a", _msgs)
|
||||
monkeypatch.setattr(opencode_service, "get_session_a", _sess)
|
||||
user = await django_user_model.objects.acreate(username="u@x.com", email="u@x.com")
|
||||
session = await ChatSession.objects.acreate(id="ses_1", user=user, is_generating=True)
|
||||
with override_settings(LANGFUSE=LF):
|
||||
await _finalize(session, TurnState("ses_1", "질문"), 1.0)
|
||||
await asyncio.gather(*list(langfuse._pending))
|
||||
assert len(lf_server) == 1
|
||||
batch = lf_server[0]["body"]["batch"]
|
||||
t, g = batch[0]["body"], batch[1]["body"]
|
||||
assert t["userId"] == "u@x.com" and t["sessionId"] == "ses_1" and t["input"] == "질문" and t["output"] == "답"
|
||||
assert g["traceId"] == t["id"] and g["usage"] == {"input": 3, "output": 5, "total": 8} and g["level"] == "DEFAULT"
|
||||
assert stream_mod is not None
|
||||
Reference in New Issue
Block a user