vincent/EnterpriseArchitect
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity
Files
474f1eddfde42b3ef2e2d3c64ac2df26531dc21d
EnterpriseArchitect/services/nvidia_sidecar
T
History
vincent 474f1eddfd fix(sidecar-v2): second-round review fixes 
- cooldown_manager: move function-level imports to module top
- proxy.py: emergency_count counter now actually increments
- server.py: metrics reads emergency_count from proxy module
- dashboard.html: real JS CDN fallback (not just comment)
- requirements.txt: remove unused prometheus_client

Round 2 review residual fixes from 沈路明/陆怀瑾/梁思筑 feedback

Co-authored-by: multica-agent <github@multica.ai>
2026-06-25 17:53:48 +08:00
..
deploy
fix(sidecar-v2): incorporate review feedback - P0/P1 fixes
2026-06-25 17:12:33 +08:00
storage
feat(sidecar-v2): implement multi-pool provider proxy with cooldown, rate limiting, WebUI
2026-06-25 16:39:01 +08:00
__init__.py
feat(sidecar-v2): implement multi-pool provider proxy with cooldown, rate limiting, WebUI
2026-06-25 16:39:01 +08:00
config.py
feat(sidecar-v2): implement multi-pool provider proxy with cooldown, rate limiting, WebUI
2026-06-25 16:39:01 +08:00
cooldown_manager.py
fix(sidecar-v2): second-round review fixes
2026-06-25 17:53:48 +08:00
crypto.py
feat(sidecar-v2): implement multi-pool provider proxy with cooldown, rate limiting, WebUI
2026-06-25 16:39:01 +08:00
dashboard.html
fix(sidecar-v2): second-round review fixes
2026-06-25 17:53:48 +08:00
docker-compose.yml
fix(sidecar-v2): incorporate review feedback - P0/P1 fixes
2026-06-25 17:12:33 +08:00
Dockerfile
fix(sidecar-v2): incorporate review feedback - P0/P1 fixes
2026-06-25 17:12:33 +08:00
main.py
feat(sidecar-v2): implement multi-pool provider proxy with cooldown, rate limiting, WebUI
2026-06-25 16:39:01 +08:00
pool_manager.py
feat(sidecar-v2): implement multi-pool provider proxy with cooldown, rate limiting, WebUI
2026-06-25 16:39:01 +08:00
proxy.py
fix(sidecar-v2): second-round review fixes
2026-06-25 17:53:48 +08:00
rate_limiter.py
feat(sidecar-v2): implement multi-pool provider proxy with cooldown, rate limiting, WebUI
2026-06-25 16:39:01 +08:00
README.md
feat(sidecar-v2): implement multi-pool provider proxy with cooldown, rate limiting, WebUI
2026-06-25 16:39:01 +08:00
requirements.txt
fix(sidecar-v2): second-round review fixes
2026-06-25 17:53:48 +08:00
router.py
feat(sidecar-v2): implement multi-pool provider proxy with cooldown, rate limiting, WebUI
2026-06-25 16:39:01 +08:00
server.py
fix(sidecar-v2): second-round review fixes
2026-06-25 17:53:48 +08:00

README.md

Sidecar V2 — Multi-Pool Provider Proxy

概述

Sidecar V2 是 OpenClaw 的 API 代理服务，实现多 Provider 池管理、负载均衡、429 冷却、RPM 队列控流。

核心功能

  • Provider 池管理：主池 (primary) + 备用池 (fallback)，支持动态增删 Provider
  • 429 冷却：检测 429 → 自动冷却 → 指数退避 → 自动恢复
  • 按 Provider 独立 RPM 限流：每个 Provider 独立的 Token Bucket
  • 路由策略：主池优先 → 备用池兜底 → 全部耗尽返 503
  • WebUI 管理：Dashboard 仪表盘 + Provider CRUD
  • 用量统计：Token 用量 + 费用统计 + 每小时/每日聚合
  • API Key 加密：AES-256-GCM 加密存储

架构

OpenClaw → Sidecar V2 (port 9190) → 路由 → 主池 Provider 1,2,3...
                                        ↘ 备池 Provider 4,5...
                                        ↘ 全部耗尽 → 503

快速开始

# 设置加密密钥 (64位十六进制)
export SIDECAR_ENCRYPTION_KEY="0000111122223333444455556666777788889999aaaabbbbccccddddeeeeffff"

# 启动服务
python3 main.py

# OR via uvicorn
python3 -m uvicorn server:app --host 127.0.0.1 --port 9190

WebUI

访问 http://127.0.0.1:9190/dashboard

API 端点

Admin API

  • GET /api/admin/backends — 列出所有 Provider
  • POST /api/admin/backends — 添加 Provider
  • PUT /api/admin/backends/{id} — 更新 Provider
  • DELETE /api/admin/backends/{id} — 删除 Provider
  • GET /api/admin/pools — 池状态汇总
  • GET /api/admin/stats/total — 总计统计
  • GET /api/admin/stats/hourly — 每小时用量
  • GET /api/admin/stats/daily — 每日聚合
  • GET /api/admin/stats/cooldown — 冷却事件历史
  • GET /api/admin/config — 系统配置

代理 API (OpenAI 兼容)

  • POST /v1/chat/completions
  • POST /v1/completions
  • POST /v1/embeddings
  • GET /v1/models

监控

  • GET /health — 健康检查
  • GET /dashboard/sse — Dashboard 实时数据流 (SSE)

环境变量

变量 默认值 说明
SIDECAR_HOST 127.0.0.1 监听地址
SIDECAR_PORT 9190 监听端口
SIDECAR_ENCRYPTION_KEY (必填) API Key 加密密钥 (64 hex chars)
SIDECAR_DB_PATH ./data/sidecar_v2.db SQLite 数据库路径
SIDECAR_RATE_RPM 40 默认 RPM 限制
SIDECAR_COOLDOWN_BASE 30 冷却基础时长 (秒)
SIDECAR_COOLDOWN_MAX 600 冷却最大时长 (秒)

存储

  • SQLite (WAL 模式)
  • 表：backends, backend_usage_logs, cooldown_events, backend_health, system_config, daily_stats
Reference in New Issue View Git Blame Copy Permalink