2f4925353a
Backend (Phase A): - A1: Adaptive silence buffer — natural_gap_ms persisted per cue; renderer computes per-cue silence_before/silence_after instead of fixed 500ms; per-cue silence files - A2: Forward-preferred snap — snap_pause_point prefers boundaries up to 4s ahead over boundaries within 1.5s behind, reducing mid-scene cuts - A3: Min-gap validation — pause points with < 200ms gap trigger forward search to the next acceptable gap - natural_gap_ms added to PausePointData model and api.ts type - New config fields: whisper_snap_forward_window, whisper_snap_backward_window, ad_silence_buffer_default, ad_silence_buffer_min_after, ad_min_acceptable_gap - Tests: test_whisper_snap.py (13 tests), test_video_renderer_buffers.py Frontend (Phase B): - B1: Drag pause-point markers — pointer state machine with 3px move threshold, clamp to min/max bounds, click-without-move still opens PausePointEditor - B2: Drag freeze blocks — orange blocks translate with linked pause point - B3: Time tooltip visible during drag, hidden on release - Tests: TimelinePreview.drag.test.tsx (10 tests) Fixes: - Share link pointed to ai-sandbox.oliver.solutions — added app_url to Settings with correct optical-dev.oliver.solutions default; share_url now configurable via APP_URL env var - Removed all ai-sandbox.oliver.solutions references from docker-compose, apache config, docs, and scripts Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
6.8 KiB
6.8 KiB
Infrastructure — Accessible Video Processing Platform
Generated: 2026-05-01
Quick Navigation
Agent Entry
| Signal | Value |
|---|---|
| Purpose | Inventory of servers, Docker services, ports, and external dependencies |
| Read When | You need to know what runs where, port mappings, or external service dependencies |
| Skip When | You need how-to deploy → runbook.md; API contracts → api_spec.md |
| Canonical | Yes |
| Next Docs | Runbook, Architecture |
| Primary Sources | docker-compose.yml, .env.example |
Server Inventory
| Server | Role | Environment |
|---|---|---|
optical-web-1 |
Production host — runs all Docker services | Production |
| Local machine | Developer workstation — Docker Compose local stack | Development |
Production URL: https://optical-dev.oliver.solutions/video-accessibility
Production API URL: https://optical-dev.oliver.solutions/video-accessibility-back
Docker Services
All services are defined in docker-compose.yml and share the accessible-video-network bridge network.
| Service | Image / Build | Container Name | Purpose |
|---|---|---|---|
mongodb |
mongo:7.0 |
accessible-video-mongodb |
Primary database |
redis |
redis:7-alpine |
accessible-video-redis |
Celery broker + result backend |
api |
./backend (target: api) |
accessible-video-api |
FastAPI REST + WebSocket |
worker |
./backend (target: worker) |
accessible-video-worker |
Celery: default, ingest, notify, render queues |
tts-worker |
./backend (target: worker) |
accessible-video-tts-worker |
Celery: tts queue |
ffmpeg-worker |
./backend (target: worker) |
accessible-video-ffmpeg-worker |
Celery: ffmpeg queue |
whisper-worker |
./backend (target: whisper-worker) |
accessible-video-whisper-worker |
Celery: whisper queue |
Port Allocation
| Service | Internal Port | External Port | Notes |
|---|---|---|---|
api |
8000 | 8012 | Exposed to host |
mongodb |
27017 | — | Internal only |
redis |
6379 | — | Internal only |
| Workers | — | — | No HTTP port |
Production: nginx reverse-proxies optical-web-1:8012 → https://optical-dev.oliver.solutions/video-accessibility-back.
Worker Configuration
| Worker | Celery Queues | Concurrency | Memory Limit | Notes |
|---|---|---|---|---|
worker |
default, ingest, notify, render |
${WORKER_CONCURRENCY:-8} |
— | General pipeline |
tts-worker |
tts |
${TTS_WORKER_CONCURRENCY:-2} |
— | Configurable via env |
ffmpeg-worker |
ffmpeg |
${FFMPEG_WORKER_CONCURRENCY:-1} |
— | CPU-bound; 1 local, higher in Cloud Run mode |
whisper-worker |
whisper |
${WHISPER_WORKER_CONCURRENCY:-1} |
8 GB (4 GB reserved) | RAM-bound; Whisper large-v3 needs ~4–6 GB |
Cloud Run offload: when FFMPEG_SERVICE_URL or WHISPER_SERVICE_URL are set, the respective workers delegate to Cloud Run HTTP endpoints instead of running locally.
Volumes
| Volume Name | Mounted In | Purpose |
|---|---|---|
accessible-video-mongodb-data |
mongodb /data/db |
MongoDB data |
accessible-video-mongodb-config |
mongodb /data/configdb |
MongoDB config |
accessible-video-redis-data |
redis /data |
Redis AOF persistence |
accessible-video-api-logs |
api /app/logs |
API log files |
accessible-video-worker-logs |
worker /app/logs |
Worker log files |
accessible-video-tts-worker-logs |
tts-worker /app/logs |
TTS worker logs |
accessible-video-ffmpeg-worker-logs |
ffmpeg-worker /app/logs |
FFmpeg worker logs |
accessible-video-whisper-worker-logs |
whisper-worker /app/logs |
Whisper worker logs |
accessible-video-shared-tmp |
worker, tts/ffmpeg/whisper workers /shared-tmp |
Shared temp dir for FFmpeg operations |
./secrets (bind mount) |
all services /secrets:ro |
GCP credentials JSON |
External Dependencies
| Service | Purpose | Config Key |
|---|---|---|
| Google Cloud Storage | Video files, VTT, MP3, rendered video | GCS_BUCKET, GOOGLE_APPLICATION_CREDENTIALS |
| MongoDB Atlas | Production database (can also run local container) | MONGODB_URI, MONGODB_DB |
| Gemini 2.5 Pro | VTT generation, translation | GEMINI_API_KEY |
| Google Cloud Translate | Language translation | TRANSLATE_API_KEY |
| Google Cloud TTS | Audio description synthesis | GOOGLE_TTS_CREDENTIALS |
| ElevenLabs | Premium TTS synthesis | ELEVENLABS_API_KEY |
| SendGrid | Transactional email (notifications, delivery) | SENDGRID_API_KEY |
| Azure AD / Microsoft MSAL | SSO authentication | AZURE_CLIENT_ID, AZURE_AUTHORITY, AZURE_REDIRECT_URI |
| Sentry | Error tracking and alerting | SENTRY_DSN |
| OpenTelemetry / OTLP | Distributed tracing | OTEL_EXPORTER_OTLP_ENDPOINT |
| AI Cost Tracker | Cross-project AI cost tracking (optical-dev) | COST_TRACKER_BASE_URL, COST_TRACKER_API_KEY |
| Google Cloud Run (optional) | Offload FFmpeg and Whisper workloads | FFMPEG_SERVICE_URL, WHISPER_SERVICE_URL |
Health Checks
| Service | Check | Interval |
|---|---|---|
mongodb |
TCP port 27017 check | 60s (10s timeout, 3 retries) |
redis |
redis-cli ping |
30s (10s timeout, 3 retries) |
api |
Depends on mongodb + redis healthy | — |
| Workers | Depend on mongodb + redis healthy | — |
Admin health endpoint: GET /api/v1/admin/health/detailed (requires admin Bearer token).
Redis Configuration
maxmemory 2gb
maxmemory-policy allkeys-lru
appendonly yes
Log Rotation
All services use Docker json-file logging:
max-size: 10m
max-file: 3
Maintenance
Last Updated: 2026-05-01
Update Triggers:
- New service added to
docker-compose.yml - Port mapping changes
- New external dependency added
- Worker queue or concurrency configuration changes
Verification:
- All services in
docker-compose.ymllisted - Port table matches
ports:sections in compose file - External dependencies match env vars in
.env.example - Volume names match
volumes:section in compose file