- Use nickfedor/watchtower:latest (maintained fork, v1.14.0)
- Add notification template for updates and failures
- Template only sends when there are actual changes
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Use list format for environment variables as suggested in shoutrrr
issue #45 to avoid YAML parsing issues with colon in bot token.
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
wget tries IPv6 [::1] when using localhost, but Node.js apps typically
listen on IPv4 only. Using 127.0.0.1 ensures IPv4 is used.
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
The /health endpoint requires a new Docker image build. Use root path
temporarily until the new image is deployed via Watchtower.
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Shoutrrr has URL parsing issues with Telegram bot tokens containing
colons. Auto-deployment works fine - notifications can be added later
via n8n webhook workflow if needed.
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Move notification URL to .env.macmini with URL-encoded colon (%3A)
in the Telegram bot token to avoid shoutrrr parsing issues.
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Switch from direct Telegram URL (which has parsing issues with colon
in bot token) to generic HTTP webhook via n8n for reliable notifications.
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
- Enable WATCHTOWER_DEBUG for verbose logs
- Add trailing slash to telegram URL (shoutrrr format fix)
- Enable WATCHTOWER_NOTIFICATION_REPORT for testing
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
New dashboards:
- Application Details: Node.js runtime (heap, event loop, GC),
HTTP details (status codes, methods, top routes), error analysis
- Database Details: PostgreSQL and Redis metrics with detailed breakdowns
Alerting rules (docker/prometheus/alerts.yml):
- Service: down, high/very high error rate, slow response time
- Infrastructure: high CPU/memory/disk usage
- Database: PostgreSQL/Redis down, high connections, low cache hit
- Container: high CPU/memory, restarts
All dashboards include service selector variable for filtering.
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
- Remove old Hetzner deployment workflows (cd-staging, cd-production)
- Remove staging docker-compose files
- Remove outdated staging/Hetzner documentation
- Add Watchtower to docker-compose.macmini.yml for auto-updates
- Update CLAUDE.md with Mac Mini server access
- Simplify docs/DEPLOYMENT.md for new architecture
Production now runs on Mac Mini with automatic deployments via Watchtower.
Co-Authored-By: Claude <noreply@anthropic.com>
- Add n8n container with PostgreSQL backend
- Configure webhook URL for n8n.mana.how
- Add persistent volume for n8n data
- Set timezone to Europe/Berlin
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
- Add MinIO service for S3-compatible object storage
- Configure contacts-backend with S3 environment variables
- Add minio_data volume
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
- Add Dockerfile for contacts-backend (port 3015)
- Add Dockerfile for contacts-web (port 5184)
- Add docker-entrypoint.sh for database migrations
- Update CI workflow with contacts-backend and contacts-web build jobs
- Add contacts services to docker-compose.macmini.yml
- Update CORS origins to include contacts.mana.how
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Add PUBLIC_TODO_BACKEND_URL and PUBLIC_CONTACTS_API_URL environment
variables to calendar-web container for cross-app integrations.
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
- CI: Build Docker images for linux/amd64 + linux/arm64
- CI: Add manacore-web to build matrix
- Add docker-compose.macmini.yml for Mac Mini deployment
- Add cloudflared-config.yml for Cloudflare Tunnel routing
- Add Mac Mini deployment scripts and documentation
- Configure Cloudflared as launchd service for auto-start
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
