refactor(docker): 优化nginx配置，修复文件上传1M的限制问题 (#48)
移除vllm模块的docker-compose配置及相关脚本，优化management-frontend的nginx配置
This commit is contained in:
zstar
GitHub
parent
83f4ed5390
commit
35bb70cee9
|
|
@ -32,10 +32,12 @@ services:
|
|||
management-frontend:
|
||||
container_name: ragflowplus-management-frontend
|
||||
image: zstar1003/ragflowplus-management-web:v0.2.0
|
||||
build:
|
||||
context: .
|
||||
dockerfile: Dockerfile
|
||||
target: frontend
|
||||
# build:
|
||||
# context: .
|
||||
# dockerfile: Dockerfile
|
||||
# target: frontend
|
||||
volumes:
|
||||
- ./nginx/management_nginx.conf:/etc/nginx/conf.d/default.conf
|
||||
ports:
|
||||
- "8888:80"
|
||||
depends_on:
|
||||
|
|
@ -44,7 +46,7 @@ services:
|
|||
- API_BASE_URL=/api
|
||||
networks:
|
||||
- ragflow
|
||||
|
||||
|
||||
# 后台信息管理系统后端
|
||||
management-backend:
|
||||
container_name: ragflowplus-management-backend
|
||||
|
|
|
|||
|
|
@ -32,10 +32,12 @@ services:
|
|||
management-frontend:
|
||||
container_name: ragflowplus-management-frontend
|
||||
image: zstar1003/ragflowplus-management-web:v0.2.0
|
||||
build:
|
||||
context: .
|
||||
dockerfile: Dockerfile
|
||||
target: frontend
|
||||
# build:
|
||||
# context: .
|
||||
# dockerfile: Dockerfile
|
||||
# target: frontend
|
||||
volumes:
|
||||
- ./nginx/management_nginx.conf:/etc/nginx/conf.d/default.conf
|
||||
ports:
|
||||
- "8888:80"
|
||||
depends_on:
|
||||
|
|
|
|||
|
|
@ -0,0 +1,26 @@
|
|||
server {
|
||||
listen 80;
|
||||
|
||||
client_max_body_size 500M;
|
||||
|
||||
location / {
|
||||
root /usr/share/nginx/html;
|
||||
try_files $uri $uri/ /index.html;
|
||||
}
|
||||
|
||||
location /v3-admin-vite/ {
|
||||
alias /usr/share/nginx/html/;
|
||||
try_files $uri $uri/ /index.html;
|
||||
}
|
||||
|
||||
location /api/ {
|
||||
# 将所有以/api/开头的请求转发到后端服务(management-backend容器的5000端口)
|
||||
proxy_pass http://management-backend:5000/api/;
|
||||
# 设置代理请求头
|
||||
proxy_set_header Host $host; # 保留原始请求的Host头
|
||||
# 传递客户端真实IP
|
||||
proxy_set_header X-Real-IP $remote_addr; # 记录客户端IP
|
||||
# 添加X-Forwarded-For头
|
||||
proxy_set_header X-Forwarded-For $proxy_add_x_forwarded_for; # 代理链路追踪
|
||||
}
|
||||
}
|
||||
|
|
@ -1,50 +0,0 @@
|
|||
services:
|
||||
vllm-bge:
|
||||
image: vllm/vllm-openai:latest
|
||||
ipc: host
|
||||
volumes:
|
||||
- ./models/bge-m3:/models
|
||||
command: [
|
||||
"--model", "/models",
|
||||
"--served-model-name", "bge-m3",
|
||||
"--dtype", "float16",
|
||||
"--gpu-memory-utilization", "0.9",
|
||||
]
|
||||
ports:
|
||||
- "8000:8000"
|
||||
deploy:
|
||||
resources:
|
||||
reservations:
|
||||
devices:
|
||||
- driver: nvidia
|
||||
capabilities: [gpu]
|
||||
networks:
|
||||
- ragflow
|
||||
|
||||
vllm-deepseek:
|
||||
image: vllm/vllm-openai:latest
|
||||
ipc: host
|
||||
volumes:
|
||||
- ./models/DeepSeek-R1-1.5B:/models
|
||||
command: [
|
||||
"--model", "/models",
|
||||
"--served-model-name", "deepseek-r1",
|
||||
"--dtype", "float16",
|
||||
"--tensor-parallel-size", "1",
|
||||
"--max-model-len", "4096"
|
||||
]
|
||||
ports:
|
||||
- "8001:8000"
|
||||
deploy:
|
||||
resources:
|
||||
reservations:
|
||||
devices:
|
||||
- driver: nvidia
|
||||
capabilities: [gpu]
|
||||
networks:
|
||||
- ragflow
|
||||
|
||||
networks:
|
||||
ragflow:
|
||||
name: docker_ragflow
|
||||
driver: bridge
|
||||
|
|
@ -1,34 +0,0 @@
|
|||
import os
|
||||
from huggingface_hub import snapshot_download
|
||||
|
||||
# 1. 设置镜像源（国内加速）
|
||||
# os.environ["HF_ENDPOINT"] = "https://mirrors.tuna.tsinghua.edu.cn/hugging-face/"
|
||||
|
||||
# 2. 定义模型列表（名称 + 下载路径）
|
||||
models_to_download = [
|
||||
{
|
||||
"repo_id": "BAAI/bge-m3", # Embedding 模型
|
||||
"local_dir": os.path.expanduser("./models/bge-m3"),
|
||||
},
|
||||
{
|
||||
"repo_id": "deepseek-ai/DeepSeek-R1-Distill-Qwen-1.5B", # LLM 模型
|
||||
"local_dir": os.path.expanduser("./models/DeepSeek-R1-1.5B"),
|
||||
}
|
||||
]
|
||||
|
||||
# 3. 遍历下载所有模型
|
||||
for model in models_to_download:
|
||||
while True: # 断点续传重试机制
|
||||
try:
|
||||
print(f"开始下载模型: {model['repo_id']} 到目录: {model['local_dir']}")
|
||||
snapshot_download(
|
||||
repo_id=model["repo_id"],
|
||||
local_dir=model["local_dir"],
|
||||
resume_download=True, # 启用断点续传
|
||||
force_download=False, # 避免重复下载已有文件
|
||||
token=None, # 如需访问私有模型，替换为你的 token
|
||||
)
|
||||
print(f"模型 {model['repo_id']} 下载完成！")
|
||||
break
|
||||
except Exception as e:
|
||||
print(f"下载失败: {e}, 重试中...")
|
||||
|
|
@ -1,49 +0,0 @@
|
|||
import requests
|
||||
from openai import OpenAI
|
||||
|
||||
# 测试 embedding 模型 (vllm-bge)
|
||||
def test_embedding(model, text):
|
||||
"""测试嵌入模型"""
|
||||
client = OpenAI(base_url="http://localhost:8000/v1", api_key="1")
|
||||
|
||||
response = client.embeddings.create(
|
||||
model=model, # 使用支持嵌入的模型
|
||||
input=text # 需要嵌入的文本
|
||||
)
|
||||
|
||||
# 打印嵌入响应内容
|
||||
# print(f"Embedding response: {response}")
|
||||
|
||||
result = response.data[0].embedding
|
||||
|
||||
if response and response.data:
|
||||
print(len(result))
|
||||
else:
|
||||
print("Failed to get embedding.")
|
||||
|
||||
# 测试文本生成模型 (vllm-deepseek)
|
||||
def test_chat(model, prompt):
|
||||
"""测试文本生成模型"""
|
||||
client = OpenAI(base_url="http://localhost:8001/v1", api_key="1")
|
||||
|
||||
response = client.completions.create(
|
||||
model=model,
|
||||
prompt=prompt
|
||||
)
|
||||
|
||||
# 打印生成的文本
|
||||
print(f"Chat response: {response.choices[0].text}")
|
||||
|
||||
def main():
|
||||
# 测试文本生成模型 deepseek-r1
|
||||
prompt = "你好，今天的天气怎么样？"
|
||||
print("Testing vllm-deepseek model for chat...")
|
||||
test_chat("deepseek-r1", prompt)
|
||||
|
||||
# 测试嵌入模型 bge-m3
|
||||
embedding_text = "我喜欢编程，尤其是做AI模型。"
|
||||
print("\nTesting vllm-bge model for embedding...")
|
||||
test_embedding("bge-m3", embedding_text)
|
||||
|
||||
if __name__ == "__main__":
|
||||
main()
|
||||
Loading…
Reference in New Issue