【云计算实战】ECS / 轻量应用服务器 + 百炼大模型 AI 服务完整部署手册｜选型规划、环境搭建、API 对接、容器部署、运维排错全套方案

#!/bin/bash
# ai_server_init.sh 云服务器AI环境初始化脚本
echo "====开始系统更新===="
sudo yum makecache
sudo yum update -y
echo "====安装基础编译依赖===="
sudo yum groupinstall -y "Development Tools"
sudo yum install -y openssl-devel bzip2-devel readline-devel sqlite-devel libffi-devel git wget curl
echo "====安装pyenv管理Python版本===="
git clone https://gitee.com/mirrors/pyenv.git ~/.pyenv
echo 'export PYENV_ROOT="$HOME/.pyenv"' >> ~/.bashrc
echo 'export PATH="$PYENV_ROOT/bin:$PATH"' >> ~/.bashrc
echo 'eval "$(pyenv init -)"' >> ~/.bashrc
source ~/.bashrc
# 安装Python3.11（AI项目推荐版本）
pyenv install 3.11.9
pyenv global 3.11.9
echo "====安装Docker与Docker‑Compose===="
curl -fsSL https://get.docker.com | bash
sudo systemctl enable docker
sudo systemctl start docker
sudo usermod -aG docker $USER
echo "====完成环境初始化，请执行 source ~/.bashrc 使环境变量生效===="
python3 --version
docker -v

import os
from http import HTTPStatus
import dashscope
from dashscope import Generation
# 从服务器环境变量读取密钥，服务器执行 export DASHSCOPE_API_KEY="sk‑xxxx"
dashscope.api_key = os.getenv("DASHSCOPE_API_KEY")
def qwen_chat(user_input: str, system_prompt: str = "你是专业AI开发助手，回答简洁专业"):
    """百炼大模型同步对话接口"""
    resp = Generation.call(
        model="qwen‑plus‑latest",
        messages=[
            {"role":"system","content":system_prompt},
            {"role":"user","content":user_input}
        ],
        result_format="message",
        stream=False,
        temperature=0.7,
        max_tokens=2048
    )
    if resp.status_code == HTTPStatus.OK:
        content = resp.output.choices[0].message.content
        input_tokens = resp.usage.input_tokens
        output_tokens = resp.usage.output_tokens
        return {
            "success":True,
            "reply":content,
            "input_tokens":input_tokens,
            "output_tokens":output_tokens
        }
    else:
        return {
            "success":False,
            "code":resp.code,
            "message":resp.message
        }
if __name__ == "__main__":
    res = qwen_chat("请简单介绍ECS部署AI服务注意事项")
    print(res)

import os
import dashscope
from dashscope import Generation
dashscope.api_key = os.getenv("DASHSCOPE_API_KEY")
def qwen_stream_chat(user_input):
    gen = Generation.call(
        model="qwen‑turbo",
        messages=[{"role":"user","content":user_input}],
        stream=True,
        incremental_output=True
    )
    for chunk in gen:
        if chunk.status_code == 200:
            yield chunk.output.choices[0].message.content
if __name__ == "__main__":
    for text in qwen_stream_chat("简述大模型部署安全要点"):
        print(text,end="")

import os
from openai import OpenAI
client = OpenAI(
    api_key=os.getenv("DASHSCOPE_API_KEY"),
    base_url="https://dashscope.aliyuncs.com/compatible-mode/v1"
)
resp = client.chat.completions.create(
    model="qwen‑plus‑latest",
    messages=[{"role":"user","content":"写一段python读取csv示例代码"}]
)
print(resp.choices[0].message.content)

import os
from flask import Flask,request,jsonify
import dashscope
from dashscope import Generation
app = Flask(__name__)
dashscope.api_key = os.getenv("DASHSCOPE_API_KEY")
def llm_call(prompt):
    rsp = Generation.call(
        model="qwen‑turbo",
        messages=[{"role":"user","content":prompt}],
        result_format="message"
    )
    if rsp.status_code == 200:
        return rsp.output.choices[0].message.content
    return f"error:{rsp.code}"
# 业务接口，简单鉴权，生产环境建议完善token认证逻辑
@app.route("/api/ai/generate",methods=["POST"])
def ai_generate():
    req_json = request.get_json()
    token = req_json.get("token","")
    if token != os.getenv("SERVER_API_TOKEN"):
        return jsonify({"code":403,"msg":"鉴权失败"}),403
    user_prompt = req_json.get("prompt","")
    result_text = llm_call(user_prompt)
    return jsonify({
        "code":200,
        "data":{"content":result_text}
    })
@app.route("/health",methods=["GET"])
def health():
    return jsonify({"status":"ok"})
if __name__ == "__main__":
    # 0.0.0.0监听全部网卡，不能写127.0.0.1，否则外网访问不到
    app.run(host="0.0.0.0",port=8090,debug=False)

# 设置环境变量，实际部署替换成自己密钥与业务token
export DASHSCOPE_API_KEY="sk‑xxx"
export SERVER_API_TOKEN="my_biz_token_123456"
# -w 工作进程数量，-b绑定地址端口
gunicorn -w 2 -b 0.0.0.0:8090 app:app

[Unit]
Description=AI‑Business‑Service
After=network.target
[Service]
User=root
WorkingDirectory=/home/ubuntu/ai_project
Environment="DASHSCOPE_API_KEY=sk‑xxx"
Environment="SERVER_API_TOKEN=my_biz_token_123456"
ExecStart=/root/.pyenv/shims/gunicorn -w 2 -b 0.0.0.0:8090 app:app
Restart=always
RestartSec=5
[Install]
WantedBy=multi‑user.target

sudo systemctl daemon‑reload
sudo systemctl enable ai‑service
sudo systemctl start ai‑service
sudo systemctl status ai‑service

FROM python:3.11‑slim
WORKDIR /app
COPY requirements.txt .
RUN pip install --no‑cache‑dir -r requirements.txt
COPY app.py .
EXPOSE 8090
CMD ["gunicorn","-w","2","-b","0.0.0.0:8090","app:app"]

flask==2.3.3
gunicorn==21.2.0
dashscope
openai

docker build -t ai‑biz‑service:v1 .
# 传入环境变量启动容器
docker run -d \
‑‑name ai‑service‑demo \
‑p 8090:8090 \
‑e DASHSCOPE_API_KEY="sk‑xxx" \
‑e SERVER_API_TOKEN="my_biz_token_123456" \
ai‑biz‑service:v1

docker run -d \
‑‑name qwen‑infer \
‑‑gpus all \
‑‑network host \
‑‑ipc=host \
‑v /data/model:/model \
lmsysorg/sglang:latest \
python3‑m sglang.launch_server \
‑‑model‑path /model/Qwen‑7B‑Instruct \
‑‑port 30000 \
‑‑max‑model‑len 16384

server {
    listen 80;
    server_name your‑domain.com;
    location / {
        proxy_pass http://127.0.0.1:8090;
        proxy_set_header Host $host;
        proxy_set_header X‑Real‑IP $remote_addr;
        proxy_read_timeout 300;
    }
}