Hilfsdienste (Monitoring, Forgejo, Glitchtip, Umami) wandern von der
auslastungs-kritischen Mac-Mini-Box auf die Windows-GPU-Box, die
ohnehin 95 % System-RAM idle hat. Production-Hot-Path bleibt auf dem
Mini, kein Geld ausgegeben, Single-Point-of-Failure am Standort
reduziert.
Stand 2026-05-06: Phase 0–2b shipped (WSL2-Docker, Grafana cross-box,
Forgejo, Umami healthy). Phase 2c (Loki+VM+Alerts) und Phase 4
(Cloudflare-Cutover für grafana.mana.how) brauchen eigene Sessions —
beides Pre-existing-Mis-config-Aufräumen, kein Architektur-Risiko.
Hardware-Inventar in WINDOWS_GPU_SERVER_SETUP.md ergänzt: Ryzen 9 5950X,
64 GB DDR4, RTX 3090, 660 GB frei C:. WSL2 auf 24 GB / 12 vCPU
gedeckelt damit AI-Scheduled-Tasks > 30 GB Reserve haben.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
mana-voice-bot's source default was 3050, which collided with mana-sync.
Today the collision is latent (voice-bot isn't deployed anywhere), but
sooner or later someone is going to start it on a host that's already
running mana-sync and the second one will refuse to bind. Moving to
3024 puts it inside the AI/ML port range alongside its dependencies
(stt 3020, tts 3022, image-gen 3023, llm 3025) and away from sync.
Updated:
- app/main.py — PORT default 3050 → 3024
- start.sh, setup.sh — same fix in the example commands
- CLAUDE.md — full rewrite. Old version described "Mac Mini deployment"
with launchd; the new version explicitly says "not deployed yet" and
documents the seven concrete steps to deploy on the Windows GPU box
alongside the other AI services (Scheduled Task, service.pyw, .env,
firewall rule, cloudflared route, WINDOWS_GPU_SERVER_SETUP.md update).
docs/WINDOWS_GPU_SERVER_SETUP.md:
- Added the missing ManaVideoGen scheduled task to all four
Start-ScheduledTask snippets — video-gen has been running on the
Windows GPU but the doc had never picked it up.
- Added a "mana-video-gen (Port 3026)" service section parallel to the
existing image-gen one, with venv path, repo pointer, model, etc.
- Added a repo-pendants table mapping C:\mana\services\<svc>\ to the
corresponding services/<svc>/ directory in the repo, plus a note that
changes should flow repo→Windows, not the other way around.
docs/PORT_SCHEMA.md:
- Reconciled the warning block with the post-cleanup reality: no more
active or latent port collisions (image-gen ↔ video-gen and
voice-bot ↔ sync are both resolved). Listed the actual ports per host
with public URLs. Kept the planned-vs-actual disclaimer for the
services that still don't match the aspirational ranges (mana-credits
3061 vs planned 3002, etc).
New GPU service for fast text-to-video generation using LTX-Video (~2B params)
on the RTX 3090. Generates 480p clips in 10-30 seconds, uses ~10GB VRAM.
Includes Cloudflare Tunnel route, Prometheus monitoring, and health checks.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
- Add API key authentication to all GPU services (X-API-Key header)
- /health and /docs remain public (no key needed)
- Shared key configured via GPU_API_KEY env variable
- Add VRAM auto-unload for mana-image-gen (5min) and mana-stt (10min)
- FLUX.2 pipeline freed after idle, recovering ~13GB VRAM
- WhisperX models freed after idle, recovering ~3GB VRAM
- Install Piper TTS voices (Thorsten + Kerstin) for local German TTS
- Update @manacore/shared-gpu client to support apiKey parameter
- Add GPU_API_KEY to .env.development
- Document API auth and VRAM management in setup guide
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Add reverse proxy entries for picture.mana.how and storage.mana.how to
Caddyfile.production. Create implementation plan and setup guide for
Windows GPU server as dedicated AI/ML node alongside Mac Mini.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
