Hitonabi
cff3f0b1a8
Backend-Services: fit/caps/sources (portiert), discover (live HF + Fit + Caps + ranked recommendation), llama-swap write/register + groups (Ko- Residenz swap:false), LiteLLM-Gateway-Config + gateway-Service (model:auto + Fallbacks). Router: discover/fit/register/groups/routing; health zeigt gateway_reachable. Frontend: Modelle&Routing mit Caps-Chips, Fit-Badges, Discover-Tab (live), Routing-View. Lokal verifiziert: Backend-Smoke (alle Endpunkte) + Frontend-Build + Browser (Shell, Discover, Caps/Fit). Box-Verifikation offen. Docs: README + docs/STATUS.md (Phasen-Tracker + Resume-Guide). Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
70 lines
2.4 KiB
Python
70 lines
2.4 KiB
Python
"""
|
|
Routing-Gateway-Service: liest/schreibt die LiteLLM-Config und prüft die
|
|
Erreichbarkeit. MC verwaltet damit die Modell-Zuordnung (welcher llama-swap-Alias
|
|
ist fast/heavy/vision/coder) und die Routing-Regeln.
|
|
"""
|
|
|
|
import httpx
|
|
|
|
from config import GATEWAY_CONFIG_PATH, GATEWAY_URL, yaml
|
|
|
|
|
|
def read_gateway_config() -> dict:
|
|
if not GATEWAY_CONFIG_PATH.exists():
|
|
return {"model_list": [], "litellm_settings": {}, "router_settings": {}}
|
|
with GATEWAY_CONFIG_PATH.open("r", encoding="utf-8") as f:
|
|
return yaml.load(f) or {}
|
|
|
|
|
|
def write_gateway_config(cfg: dict) -> None:
|
|
import os
|
|
GATEWAY_CONFIG_PATH.parent.mkdir(parents=True, exist_ok=True)
|
|
tmp = GATEWAY_CONFIG_PATH.with_name(GATEWAY_CONFIG_PATH.name + ".tmp")
|
|
with tmp.open("w", encoding="utf-8") as f:
|
|
yaml.dump(cfg, f)
|
|
os.replace(tmp, GATEWAY_CONFIG_PATH)
|
|
|
|
|
|
def routing_summary() -> dict:
|
|
"""Kompakte Sicht für die UI: welcher Backend-Alias steckt hinter welchem
|
|
Gateway-Modellnamen + die Fallback-Ketten."""
|
|
cfg = read_gateway_config()
|
|
routes = []
|
|
for entry in cfg.get("model_list") or []:
|
|
params = entry.get("litellm_params") or {}
|
|
routes.append({
|
|
"name": entry.get("model_name"),
|
|
"target": str(params.get("model", "")),
|
|
"api_base": params.get("api_base"),
|
|
})
|
|
settings = cfg.get("litellm_settings") or {}
|
|
return {
|
|
"routes": routes,
|
|
"fallbacks": settings.get("fallbacks") or [],
|
|
"context_window_fallbacks": settings.get("context_window_fallbacks") or [],
|
|
}
|
|
|
|
|
|
def set_route(name: str, target_alias: str, api_base: str = "http://127.0.0.1:8080/v1") -> None:
|
|
"""Einen Gateway-Modellnamen (z.B. 'fast') auf einen llama-swap-Alias mappen."""
|
|
cfg = read_gateway_config()
|
|
ml = cfg.setdefault("model_list", [])
|
|
params = {"model": f"openai/{target_alias}", "api_base": api_base, "api_key": "sk-noauth"}
|
|
for entry in ml:
|
|
if entry.get("model_name") == name:
|
|
entry["litellm_params"] = params
|
|
break
|
|
else:
|
|
ml.append({"model_name": name, "litellm_params": params})
|
|
write_gateway_config(cfg)
|
|
|
|
|
|
def gateway_reachable() -> bool:
|
|
try:
|
|
with httpx.Client(timeout=3.0) as c:
|
|
# LiteLLM hat /health/liveliness; /v1/models tut's auch.
|
|
r = c.get(f"{GATEWAY_URL}/v1/models")
|
|
return r.status_code in (200, 401)
|
|
except Exception:
|
|
return False
|