PyTorch 2.8镜像部署教程：支持JWT鉴权的企业级AI API服务配置-平芜编程栈

PyTorch 2.8镜像部署教程：支持JWT鉴权的企业级AI API服务配置

1. 环境准备与快速部署

在开始之前，请确保您的服务器满足以下硬件要求：

显卡：RTX 4090D 24GB显存
内存：≥120GB
存储：系统盘50GB + 数据盘40GB

1.1 镜像拉取与启动

使用以下命令拉取并启动PyTorch 2.8镜像：

docker pull pytorch/pytorch:2.8-cuda12.4-cudnn8-devel docker run -it --gpus all -p 8000:8000 -v /data:/data pytorch/pytorch:2.8-cuda12.4-cudnn8-devel

1.2 环境验证

验证GPU是否可用：

python -c "import torch; print('PyTorch:', torch.__version__); print('CUDA available:', torch.cuda.is_available()); print('GPU count:', torch.cuda.device_count())"

预期输出应显示CUDA可用且检测到GPU设备。

2. API服务基础配置

2.1 安装必要依赖

pip install fastapi uvicorn python-jose[cryptography] passlib[bcrypt] python-multipart

2.2 创建基础API服务

新建main.py文件：

from fastapi import FastAPI app = FastAPI() @app.get("/") def read_root(): return {"message": "PyTorch 2.8 API服务已启动"}

启动服务：

uvicorn main:app --host 0.0.0.0 --port 8000

3. JWT鉴权系统实现

3.1 用户认证模块

创建auth.py文件：

from datetime import datetime, timedelta from typing import Optional from jose import JWTError, jwt from passlib.context import CryptContext SECRET_KEY = "your-secret-key" ALGORITHM = "HS256" ACCESS_TOKEN_EXPIRE_MINUTES = 30 pwd_context = CryptContext(schemes=["bcrypt"], deprecated="auto") def verify_password(plain_password: str, hashed_password: str): return pwd_context.verify(plain_password, hashed_password) def get_password_hash(password: str): return pwd_context.hash(password) def create_access_token(data: dict, expires_delta: Optional[timedelta] = None): to_encode = data.copy() if expires_delta: expire = datetime.utcnow() + expires_delta else: expire = datetime.utcnow() + timedelta(minutes=15) to_encode.update({"exp": expire}) encoded_jwt = jwt.encode(to_encode, SECRET_KEY, algorithm=ALGORITHM) return encoded_jwt

3.2 保护API端点

更新main.py：

from fastapi import Depends, FastAPI, HTTPException, status from fastapi.security import OAuth2PasswordBearer, OAuth2PasswordRequestForm from auth import * oauth2_scheme = OAuth2PasswordBearer(tokenUrl="token") @app.post("/token") async def login(form_data: OAuth2PasswordRequestForm = Depends()): # 这里应查询数据库验证用户 # 简化示例使用固定用户 if form_data.username != "admin" or not verify_password(form_data.password, get_password_hash("admin123")): raise HTTPException( status_code=status.HTTP_401_UNAUTHORIZED, detail="用户名或密码错误", headers={"WWW-Authenticate": "Bearer"}, ) access_token = create_access_token( data={"sub": form_data.username} ) return {"access_token": access_token, "token_type": "bearer"} @app.get("/protected") async def protected_route(token: str = Depends(oauth2_scheme)): try: payload = jwt.decode(token, SECRET_KEY, algorithms=[ALGORITHM]) username: str = payload.get("sub") if username is None: raise HTTPException( status_code=status.HTTP_401_UNAUTHORIZED, detail="无效认证凭证", headers={"WWW-Authenticate": "Bearer"}, ) except JWTError: raise HTTPException( status_code=status.HTTP_401_UNAUTHORIZED, detail="无效认证凭证", headers={"WWW-Authenticate": "Bearer"}, ) return {"message": "访问受保护路由成功", "username": username}

4. 模型推理服务集成

4.1 添加模型推理端点

扩展main.py：

import torch from torch import nn from fastapi import File, UploadFile class SimpleModel(nn.Module): def __init__(self): super().__init__() self.linear = nn.Linear(10, 1) def forward(self, x): return self.linear(x) model = SimpleModel().cuda() @app.post("/predict") async def predict(data: list, token: str = Depends(oauth2_scheme)): try: input_tensor = torch.tensor(data, dtype=torch.float32).cuda() with torch.no_grad(): output = model(input_tensor) return {"prediction": output.cpu().numpy().tolist()} except Exception as e: raise HTTPException(status_code=400, detail=str(e))

4.2 文件上传处理

添加文件上传端点：

@app.post("/upload") async def upload_file(file: UploadFile = File(...), token: str = Depends(oauth2_scheme)): try: contents = await file.read() # 这里可以添加文件处理逻辑 return {"filename": file.filename, "size": len(contents)} except Exception as e: raise HTTPException(status_code=400, detail=str(e))

5. 生产环境优化

5.1 使用Gunicorn多进程

安装Gunicorn：

pip install gunicorn

创建启动脚本start.sh：

#!/bin/bash gunicorn -w 4 -k uvicorn.workers.UvicornWorker -b 0.0.0.0:8000 main:app

5.2 配置Nginx反向代理

示例Nginx配置：

server { listen 80; server_name your-domain.com; location / { proxy_pass http://localhost:8000; proxy_set_header Host $host; proxy_set_header X-Real-IP $remote_addr; proxy_set_header X-Forwarded-For $proxy_add_x_forwarded_for; } }

5.3 日志与监控

配置日志记录：

import logging from fastapi.logger import logger logging.basicConfig( level=logging.INFO, format="%(asctime)s - %(name)s - %(levelname)s - %(message)s", handlers=[ logging.FileHandler("api.log"), logging.StreamHandler() ] ) logger = logging.getLogger(__name__)