hyungi/tk-factory-services
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

feat: ai-service를 ds923에서 맥미니로 이전

Browse Source 
- ChromaDB → Qdrant 전환 (맥미니 기존 인스턴스, tk_qc_issues 컬렉션)
- Ollama 임베딩/텍스트 생성 URL 분리 (임베딩: 맥미니, 텍스트: GPU서버)
- MLX fallback 제거, Ollama 단일 경로로 단순화
- ds923 docker-compose에서 ai-service 제거
- gateway/system3-web nginx: ai-service 프록시를 ai.hyungi.net 경유로 변경
- resolver + 변수 기반 proxy_pass로 런타임 DNS 해석 (컨테이너 시작 실패 방지)

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
This commit is contained in:
Hyungi Ahn
2026-03-11 15:07:58 +09:00
parent 2d25d54589
commit 85f674c9cb
9 changed files with 125 additions and 130 deletions
Download Patch File Download Diff File Expand all files Collapse all files

58
ai-service/services/ollama_client.py
View File

@@ -5,7 +5,8 @@ from config import settings
class OllamaClient:
def __init__(self):
self.base_url = settings.OLLAMA_BASE_URL
self.text_url = settings.OLLAMA_BASE_URL # GPU서버 (텍스트 생성)
self.embed_url = settings.OLLAMA_EMBED_URL # 맥미니 (임베딩)
self.timeout = httpx.Timeout(float(settings.OLLAMA_TIMEOUT), connect=10.0)
self._client: httpx.AsyncClient | None = None
@@ -22,7 +23,7 @@ class OllamaClient:
async def generate_embedding(self, text: str) -> list[float]:
client = await self._get_client()
response = await client.post(
f"{self.base_url}/api/embeddings",
f"{self.embed_url}/api/embeddings",
json={"model": settings.OLLAMA_EMBED_MODEL, "prompt": text},
)
response.raise_for_status()
@@ -43,49 +44,38 @@ class OllamaClient:
messages.append({"role": "system", "content": system})
messages.append({"role": "user", "content": prompt})
client = await self._get_client()
# 조립컴 Ollama 메인, MLX fallback
try:
response = await client.post(
f"{self.base_url}/api/chat",
json={
"model": settings.OLLAMA_TEXT_MODEL,
"messages": messages,
"stream": False,
"think": False,
"options": {"temperature": 0.3, "num_predict": 2048},
},
)
response.raise_for_status()
return response.json()["message"]["content"]
except Exception:
response = await client.post(
f"{settings.MLX_BASE_URL}/chat/completions",
json={
"model": settings.MLX_TEXT_MODEL,
"messages": messages,
"max_tokens": 2048,
"temperature": 0.3,
},
)
response.raise_for_status()
return response.json()["choices"][0]["message"]["content"]
response = await client.post(
f"{self.text_url}/api/chat",
json={
"model": settings.OLLAMA_TEXT_MODEL,
"messages": messages,
"stream": False,
"think": False,
"options": {"temperature": 0.3, "num_predict": 2048},
},
)
response.raise_for_status()
return response.json()["message"]["content"]
async def check_health(self) -> dict:
result = {}
short_timeout = httpx.Timeout(5.0, connect=3.0)
# GPU서버 Ollama (텍스트 생성)
try:
async with httpx.AsyncClient(timeout=short_timeout) as c:
response = await c.get(f"{self.base_url}/api/tags")
response = await c.get(f"{self.text_url}/api/tags")
models = response.json().get("models", [])
result["ollama"] = {"status": "connected", "models": [m["name"] for m in models]}
result["ollama_text"] = {"status": "connected", "url": self.text_url, "models": [m["name"] for m in models]}
except Exception:
result["ollama"] = {"status": "disconnected"}
result["ollama_text"] = {"status": "disconnected", "url": self.text_url}
# 맥미니 Ollama (임베딩)
try:
async with httpx.AsyncClient(timeout=short_timeout) as c:
response = await c.get(f"{settings.MLX_BASE_URL}/health")
result["mlx"] = {"status": "connected", "model": settings.MLX_TEXT_MODEL}
response = await c.get(f"{self.embed_url}/api/tags")
models = response.json().get("models", [])
result["ollama_embed"] = {"status": "connected", "url": self.embed_url, "models": [m["name"] for m in models]}
except Exception:
result["mlx"] = {"status": "disconnected"}
result["ollama_embed"] = {"status": "disconnected", "url": self.embed_url}
return result