Skip to main content

Distributed worker for HuggingFace Space scheduling

Project description

HFS v2 - HuggingFace Space Worker 分布式调度系统

基于 Redis 的分布式 Worker 调度系统,用于管理 HuggingFace Space 资源池。

特性

  • 分布式调度 - 多 Worker 并发,自动负载均衡
  • 状态管理 - 原子操作(Lua 脚本),保证一致性
  • 健康检查 - 自动检测崩溃、超时、孤儿资源
  • 账号管理 - 多账号池,自动选择、cooldown、评分
  • Space 轮换 - 自动创建、绑定、轮换、复用
  • Admin CLI - 命令行管理工具

快速开始

1. 安装依赖

pip install redis huggingface-hub click tabulate

2. 初始化系统

cd v2
python admin/init.py

3. 查看状态

python admin/cli.py --redis-url="redis://..." list-nodes
python admin/cli.py --redis-url="redis://..." list-accounts

4. 运行 Worker

python -m hfs --redis-url="redis://..." --space-id=my-space --project-id=demo --node-id=node-1

架构

┌─────────────┐
│   Redis     │  ← 状态存储
└──────┬──────┘
       │
   ┌───┴────┐
   │        │
┌──▼──┐  ┌─▼───┐
│Worker│  │Worker│  ← 独立进程
└──┬──┘  └─┬───┘
   │       │
┌──▼───────▼──┐
│  Scheduler  │  ← 调度器
└─────────────┘

核心模块

  • state.py - 状态机 + 原子操作(Lua 脚本)
  • health.py - 健康检查(崩溃检测、一致性验证)
  • policy.py - 策略配置(场景、命名)
  • worker.py - Worker 主循环(心跳、进程管理)
  • scheduler.py - 调度器(分配、轮换、创建)
  • account.py - 账号管理(选择、cooldown、评分)
  • hf.py - HuggingFace API 封装

测试

# 运行所有测试
pytest tests/ -v

# 运行特定模块
pytest tests/test_state.py -v
pytest tests/test_worker.py -v
pytest tests/test_scheduler.py -v

文档

配置

Redis

export HFS_REDIS_URL="redis://:password@host:port/db"

HuggingFace 账号

admin/init.py 中配置账号列表:

ACCOUNTS = [
    {'username': 'user1', 'token': 'hf_xxx'},
    {'username': 'user2', 'token': 'hf_yyy'}
]

开发

测试驱动开发

  1. 先写测试(tests/test_*.py
  2. 再实现功能(hfs/*.py
  3. 运行测试验证

代码结构

v2/
├── hfs/                # Worker 包
│   ├── state.py        # 状态机
│   ├── health.py       # 健康检查
│   ├── policy.py       # 策略配置
│   ├── worker.py       # Worker 主循环
│   ├── scheduler.py    # 调度器
│   ├── account.py      # 账号管理
│   └── hf.py           # HF API
├── admin/              # Admin 工具
│   ├── cli.py          # 命令行工具
│   └── init.py         # 快速初始化
├── tests/              # 测试
└── docs/               # 文档

许可

MIT License

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

mpool_worker-0.1.0.tar.gz (33.5 kB view details)

Uploaded Source

Built Distribution

If you're not sure about the file name format, learn more about wheel file names.

mpool_worker-0.1.0-py3-none-any.whl (31.8 kB view details)

Uploaded Python 3

File details

Details for the file mpool_worker-0.1.0.tar.gz.

File metadata

  • Download URL: mpool_worker-0.1.0.tar.gz
  • Upload date:
  • Size: 33.5 kB
  • Tags: Source
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/6.2.0 CPython/3.12.11

File hashes

Hashes for mpool_worker-0.1.0.tar.gz
Algorithm Hash digest
SHA256 a0ffa2f7a3c47e837ef8f0f4a04ae7579af2cc6695e8de5fdddc1db178be677c
MD5 9553c404484f497248badbea38a5b027
BLAKE2b-256 c5e76dc1af280b58aaee0b5b81149bdb675ea3343c7ce3104b27c025e50b8e02

See more details on using hashes here.

File details

Details for the file mpool_worker-0.1.0-py3-none-any.whl.

File metadata

  • Download URL: mpool_worker-0.1.0-py3-none-any.whl
  • Upload date:
  • Size: 31.8 kB
  • Tags: Python 3
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/6.2.0 CPython/3.12.11

File hashes

Hashes for mpool_worker-0.1.0-py3-none-any.whl
Algorithm Hash digest
SHA256 81a1ed3bd8dace71d45b48b56cbf8b43a4678ad673ce3de3a0f5c4fb938648d6
MD5 90b49554f8d4f8ef908049d410196c19
BLAKE2b-256 4597d2c7b9f7a9ba1fe44c71d0bbc920b957a2d79feac6d7ae68e44c062c2b70

See more details on using hashes here.

Supported by

AWS Cloud computing and Security Sponsor Datadog Monitoring Depot Continuous Integration Fastly CDN Google Download Analytics Pingdom Monitoring Sentry Error logging StatusPage Status page