fix(gpu): #1978 MuseTalk v1.5 推理修复与Worker稳定性补丁 #2003

Closed
auto-approve-bot wants to merge 0 commits from fix/gpu-worker-musetalk-patches into develop
Collaborator

变更内容

musetalk_server.py(新增文件)

  • v3 API兼容性:支持新的请求格式(audio_url/video_url字段)
  • v4 正确pipeline:使用vae.get_latents_for_unet()生成8ch输入(masked_latents+ref_latents拼接)
  • v4.4 VRAM优化:batch_size=2 + CPU offload + PYTORCH_CUDA_ALLOC_CONF=max_split_size_mb:128
  • RTX2060 6GB可稳定运行,22s视频推理约204s(0.11x实时)

gpu_worker.py

  • 健康检查响应格式适配新musetalk_server
  • REQUEST_TIMEOUT提升到900s,与服务端GPU_TASK_TIMEOUT_SECONDS对齐
  • TASK_MAX_RETRY降到1,避免与服务端MAX_ATTEMPTS=3相乘放大重试
  • 增加TASK_HEARTBEAT_INTERVAL=30s,推理期间独立线程心跳防超时
  • 增加MIN_VIDEO_DURATION_SECONDS=3s前置拦截,避免1s视频division by zero

测试

  • ✅ RTX2060 6GB 22s视频推理成功(sun_musetalk_result.mp4)
  • ⚠️ 待验证:端到端staging API链路测试

风险

  • gpu-poll.timer已临时禁用,需合入后重新启用自动部署
## 变更内容 ### musetalk_server.py(新增文件) - v3 API兼容性:支持新的请求格式(audio_url/video_url字段) - v4 正确pipeline:使用vae.get_latents_for_unet()生成8ch输入(masked_latents+ref_latents拼接) - v4.4 VRAM优化:batch_size=2 + CPU offload + PYTORCH_CUDA_ALLOC_CONF=max_split_size_mb:128 - RTX2060 6GB可稳定运行,22s视频推理约204s(0.11x实时) ### gpu_worker.py - 健康检查响应格式适配新musetalk_server - REQUEST_TIMEOUT提升到900s,与服务端GPU_TASK_TIMEOUT_SECONDS对齐 - TASK_MAX_RETRY降到1,避免与服务端MAX_ATTEMPTS=3相乘放大重试 - 增加TASK_HEARTBEAT_INTERVAL=30s,推理期间独立线程心跳防超时 - 增加MIN_VIDEO_DURATION_SECONDS=3s前置拦截,避免1s视频division by zero ## 测试 - ✅ RTX2060 6GB 22s视频推理成功(sun_musetalk_result.mp4) - ⚠️ 待验证:端到端staging API链路测试 ## 风险 - gpu-poll.timer已临时禁用,需合入后重新启用自动部署
auto-approve-bot added 1 commit 2026-09-20 21:37:07 +08:00
fix(gpu): #1978 MuseTalk v1.5 推理修复与Worker稳定性补丁
CI/CD Pipeline / Check push changed paths (pull_request) Has been skipped
CI/CD Pipeline / Dedup Check - skip PR tests when covered by push pipeline (pull_request) Successful in 3s
CI/CD Pipeline / Check if frontend-only change (pull_request) Successful in 2s
CI/CD Pipeline / Build Staging API Image (pull_request) Has been skipped
CI/CD Pipeline / Build Staging Web Image (pull_request) Has been skipped
CI/CD Pipeline / Build Staging Worker Image (pull_request) Has been skipped
CI/CD Pipeline / Frontend Lint (pull_request) Has been skipped
CI/CD Pipeline / Frontend Unit Tests (pull_request) Has been skipped
CI/CD Pipeline / PR Build API Image (pull_request) Successful in 1m48s
CI/CD Pipeline / PR Build Web Image (pull_request) Has been skipped
PR Automation / Auto Approve on CI Green (pull_request) Successful in 3m4s
CI/CD Pipeline / Retag skipped Staging API Image (pull_request) Has been skipped
CI/CD Pipeline / Retag skipped Staging Web Image (pull_request) Has been skipped
CI/CD Pipeline / Retag skipped Staging Worker Image (pull_request) Has been skipped
CI/CD Pipeline / Deploy Staging (Watchtower auto-deploy) (pull_request) Has been skipped
CI/CD Pipeline / Staging E2E Tests (pull_request) Has been skipped
CI/CD Pipeline / Staging API Integration Tests (pull_request) Has been skipped
CI/CD Pipeline / ACR Image Cleanup (pull_request) Has been skipped
CI/CD Pipeline / PR Build Worker Image (pull_request) Successful in 2m27s
AI Code Review / AI Code Review (pull_request) Successful in 7m54s
Preview Deploy / Deploy Preview Environment (pull_request) Successful in 8m56s
PR Automation / Auto Merge on CI Green + Approved (pull_request) Successful in 10m41s
CI/CD Pipeline / Validate - Style (pull_request) Successful in 39m59s
CI/CD Pipeline / Integration Tests (pull_request) Successful in 40m7s
CI/CD Pipeline / Unit Tests (pull_request) Successful in 44m57s
CI/CD Pipeline / Validate - Python (mypy + alembic) (pull_request) Successful in 1h5m19s
CI/CD Pipeline / Build Production API Image (pull_request) Has been cancelled
CI/CD Pipeline / Build Production Web Image (pull_request) Has been cancelled
CI/CD Pipeline / Build Production Worker Image (pull_request) Has been cancelled
CI/CD Pipeline / Deploy Production (pull_request) Has been cancelled
CI/CD Pipeline / Production Browser E2E (pull_request) Has been cancelled
CI/CD Pipeline / Canary Release to Production (pull_request) Has been cancelled
CI/CD Pipeline / CI Gate (pull_request) Has been cancelled
CI/CD Pipeline / Validate - Security (pull_request) Has been cancelled
e7bf85ca86
musetalk_server.py:
- v3 API兼容性:支持新的请求格式(audio_url/video_url字段)
- v4 正确pipeline:使用vae.get_latents_for_unet()生成8ch输入
- v4.4 VRAM优化:batch_size=2 + CPU offload + PYTORCH_CUDA_ALLOC_CONF=max_split_size_mb:128
- RTX2060 6GB可稳定运行,22s视频推理约204s(0.11x实时)

gpu_worker.py:
- 修复健康检查响应格式适配musetalk_server
- REQUEST_TIMEOUT提升到900s,与服务端GPU_TASK_TIMEOUT_SECONDS对齐
- TASK_MAX_RETRY降到1,避免与服务端MAX_ATTEMPTS=3相乘放大重试
- 增加TASK_HEARTBEAT_INTERVAL=30s,推理期间独立线程心跳防超时
- 增加MIN_VIDEO_DURATION_SECONDS=3s前置拦截,避免1s视频division by zero

🚀 预览环境已部署

项目 详情
PR号 #2003
预览链接 https://pr-2003.preview.xiaoxiajianji.com
API环境 staging

💡 预览环境使用 staging API 数据,请勿在预览环境中操作重要数据。

🔄 每次提交新代码后预览环境会自动更新。

🗑️ PR 关闭或合并后,预览环境会自动清理。

🚀 **预览环境已部署** | 项目 | 详情 | |------|------| | PR号 | #2003 | | 预览链接 | [https://pr-2003.preview.xiaoxiajianji.com](https://pr-2003.preview.xiaoxiajianji.com) | | API环境 | staging | > 💡 预览环境使用 staging API 数据,请勿在预览环境中操作重要数据。 > > 🔄 每次提交新代码后预览环境会自动更新。 > > 🗑️ PR 关闭或合并后,预览环境会自动清理。
auto-approve-bot added 1 commit 2026-09-20 22:51:05 +08:00
feat(gpu): add GFPGAN face enhancement + ffmpeg pipe optimization
CI/CD Pipeline / Dedup Check - skip PR tests when covered by push pipeline (pull_request) Successful in 1s
CI/CD Pipeline / Check if frontend-only change (pull_request) Successful in 1s
CI/CD Pipeline / Check push changed paths (pull_request) Has been skipped
PR Automation / Auto Approve on CI Green (pull_request) Successful in 2m43s
AI Code Review / AI Code Review (pull_request) Successful in 7m33s
Preview Deploy / Deploy Preview Environment (pull_request) Successful in 6m26s
CI/CD Pipeline / Frontend Lint (pull_request) Has been skipped
CI/CD Pipeline / Frontend Unit Tests (pull_request) Has been skipped
CI/CD Pipeline / PR Build API Image (pull_request) Successful in 15s
CI/CD Pipeline / PR Build Web Image (pull_request) Has been skipped
CI/CD Pipeline / PR Build Worker Image (pull_request) Successful in 10s
CI/CD Pipeline / Build Staging API Image (pull_request) Has been skipped
CI/CD Pipeline / Build Staging Web Image (pull_request) Has been skipped
CI/CD Pipeline / Build Staging Worker Image (pull_request) Has been skipped
CI/CD Pipeline / Retag skipped Staging API Image (pull_request) Has been skipped
CI/CD Pipeline / Retag skipped Staging Web Image (pull_request) Has been skipped
CI/CD Pipeline / Retag skipped Staging Worker Image (pull_request) Has been skipped
PR Automation / Auto Merge on CI Green + Approved (pull_request) Successful in 10m40s
CI/CD Pipeline / Deploy Staging (Watchtower auto-deploy) (pull_request) Has been skipped
CI/CD Pipeline / Validate - Style (pull_request) Successful in 22m19s
CI/CD Pipeline / Integration Tests (pull_request) Successful in 18m23s
CI/CD Pipeline / Staging E2E Tests (pull_request) Has been skipped
CI/CD Pipeline / Staging API Integration Tests (pull_request) Has been skipped
CI/CD Pipeline / ACR Image Cleanup (pull_request) Has been skipped
CI/CD Pipeline / Unit Tests (pull_request) Successful in 26m31s
CI/CD Pipeline / Validate - Python (mypy + alembic) (pull_request) Successful in 29m43s
ACR Cleanup / ACR Image Cleanup (pull_request_target) Successful in 10s
Preview Cleanup / Cleanup Preview Environment (pull_request) Successful in 7m54s
CI/CD Pipeline / Validate - Security (pull_request) Successful in 1h28m58s
CI/CD Pipeline / Build Production API Image (pull_request) Has been skipped
CI/CD Pipeline / Build Production Web Image (pull_request) Has been skipped
CI/CD Pipeline / Build Production Worker Image (pull_request) Has been skipped
CI/CD Pipeline / CI Gate (pull_request) Successful in 1s
CI/CD Pipeline / Deploy Production (pull_request) Has been skipped
CI/CD Pipeline / Canary Release to Production (pull_request) Has been skipped
CI/CD Pipeline / Production Browser E2E (pull_request) Has been skipped
93cb3e12a0
- Add GFPGANv1.4 face super-resolution (FP16, +170MB VRAM, ~64ms/frame)
  - Controllable via MUSE_USE_GFPGAN env var (default enabled)
  - Enhances generated face region before blending with original frame
- Replace PNG disk I/O with ffmpeg stdin pipe encoding (~2s faster)
  - Frames streamed directly to ffmpeg as raw BGR24, no temp files
  - Eliminates output_frames_dir/ PNG write/read cycle
- Total warm inference: 54s for 5s video (was 56s with PNG, 48s without GFPGAN)
- VRAM usage: ~3.3GB with MuseTalk+GFPGAN loaded (RTX 2060 6GB OK)
auto-approve-bot closed this pull request 2026-09-21 00:11:12 +08:00

🗑️ 预览环境已清理

PR #2003 已关闭或合并,对应的预览环境已被清理。

如有需要,可以重新打开 PR 来重新生成预览环境。

🗑️ **预览环境已清理** PR #2003 已关闭或合并,对应的预览环境已被清理。 > 如有需要,可以重新打开 PR 来重新生成预览环境。
Some checks are pending
CI/CD Pipeline / Dedup Check - skip PR tests when covered by push pipeline (pull_request) Successful in 1s
CI/CD Pipeline / Check if frontend-only change (pull_request) Successful in 1s
CI/CD Pipeline / Check push changed paths (pull_request) Has been skipped
PR Automation / Auto Approve on CI Green (pull_request) Successful in 2m43s
AI Code Review / AI Code Review (pull_request) Successful in 7m33s
Preview Deploy / Deploy Preview Environment (pull_request) Successful in 6m26s
CI/CD Pipeline / Frontend Lint (pull_request) Has been skipped
CI/CD Pipeline / Frontend Unit Tests (pull_request) Has been skipped
CI/CD Pipeline / PR Build API Image (pull_request) Successful in 15s
CI/CD Pipeline / PR Build Web Image (pull_request) Has been skipped
CI/CD Pipeline / PR Build Worker Image (pull_request) Successful in 10s
CI/CD Pipeline / Build Staging API Image (pull_request) Has been skipped
CI/CD Pipeline / Build Staging Web Image (pull_request) Has been skipped
CI/CD Pipeline / Build Staging Worker Image (pull_request) Has been skipped
CI/CD Pipeline / Retag skipped Staging API Image (pull_request) Has been skipped
CI/CD Pipeline / Retag skipped Staging Web Image (pull_request) Has been skipped
CI/CD Pipeline / Retag skipped Staging Worker Image (pull_request) Has been skipped
PR Automation / Auto Merge on CI Green + Approved (pull_request) Successful in 10m40s
CI/CD Pipeline / Deploy Staging (Watchtower auto-deploy) (pull_request) Has been skipped
CI/CD Pipeline / Validate - Style (pull_request) Successful in 22m19s
CI/CD Pipeline / Integration Tests (pull_request) Successful in 18m23s
CI/CD Pipeline / Staging E2E Tests (pull_request) Has been skipped
CI/CD Pipeline / Staging API Integration Tests (pull_request) Has been skipped
CI/CD Pipeline / ACR Image Cleanup (pull_request) Has been skipped
CI/CD Pipeline / Unit Tests (pull_request) Successful in 26m31s
CI/CD Pipeline / Validate - Python (mypy + alembic) (pull_request) Successful in 29m43s
ACR Cleanup / ACR Image Cleanup (pull_request_target) Successful in 10s
Preview Cleanup / Cleanup Preview Environment (pull_request) Successful in 7m54s
CI/CD Pipeline / Validate - Security (pull_request) Successful in 1h28m58s
CI/CD Pipeline / Build Production API Image (pull_request) Has been skipped
CI/CD Pipeline / Build Production Web Image (pull_request) Has been skipped
CI/CD Pipeline / Build Production Worker Image (pull_request) Has been skipped
CI/CD Pipeline / CI Gate (pull_request) Successful in 1s
CI/CD Pipeline / Deploy Production (pull_request) Has been skipped
CI/CD Pipeline / Canary Release to Production (pull_request) Has been skipped
CI/CD Pipeline / Production Browser E2E (pull_request) Has been skipped

Pull request closed

Sign in to join this conversation.