一、环境配置

1、进行ollama安装

# 下载官方安装脚本
curl -L https://ollama.ai/install.sh | sh

# 启动服务并设置开机自启
sudo systemctl enable --now ollama

2、安装nginx

sudo apt update && sudo apt install nginx -y
sudo systemctl enable --now nginx

二、服务验证（可选）

1、验证Ollama运行

curl http://127.0.0.1:11434
# 预期返回：Ollama is running

2、验证Nginx运行

curl -I http://localhost

三、🚀 极简三步配置

【这里可以先把ollama默认启动端口也进行修改掉，避免直接扫描11434】

【注意，ollama启动时候host设置为127.0.0.1，不要设置为0.0.0.0，否则公网仍可访问】

1. 创建Nginx配置文件（直接覆盖原有配置）

sudo tee /etc/nginx/conf.d/ollama.conf <<'EOF'
server {
    listen 11434;

    location / {
        if ($http_authorization != "Bearer funhpc-666") {  # ← 这里改成你的密码
            return 403;
        }
        proxy_pass http://127.0.0.1:11434;
    }
}
EOF
sudo nginx -t && sudo systemctl reload nginx

2. 生效配置

sudo nginx -t && sudo systemctl reload nginx
nginx -t && systemctl reload nginx

3. 测试访问

# 正确请求（记得替换密码）
curl -H "Authorization: Bearer my-secret-key-123" http://localhost:9180
# 错误测试
curl http://localhost:9180  # 应显示403错误

四、测试请求

1、Crul测试案例

curl -H "Authorization: Bearer my-secret-key-123" \
  -X POST http://https://riq853u87qswxqntalk4090.funhpc.com:30499/api/generate \
  -d '{
    "model": "deepseek-r1:1.5b",
    "prompt": "为什么天空是蓝色的？",
    "stream": false
  }'

2、OpenAI调用效果

from openai import OpenAI
client = OpenAI(api_key="my-secret-key-123", base_url="https://riq853u87qswxqntalk4090.funhpc.com:30499/v1")

messages = [
    {"role": "system", "content": "You are a helpful assistant"},
    {"role": "user", "content": "最厉害的一个奥特曼是谁？"}
]

# 调用Chat Completions接口（必填model和messages）
response = client.chat.completions.create(
    model="deepseek-r1:1.5b",  # 需确认后端是否支持该模型名称
    messages=messages,
    stream=True)
# print(response)
reasoning_content = ""
content = ""

for chunk in response:
    delta = chunk.choices[0].delta
    # print(chunk)
    if hasattr(delta, "reasoning_content") and delta.reasoning_content:
        reasoning_content += delta.reasoning_content
        print(f"{delta.reasoning_content}", end="", flush=True)
    if hasattr(delta, "content") and delta.content:
        content += delta.content
        print(f"{delta.content}", end="", flush=True)