- Gitea Actions workflow now syncs repo to server, builds images, restarts containers, and runs health checks
- Removed all hardcoded secrets from scripts/deploy.sh
- Added CI/CD documentation and ignored .env.deploy
NOTE: Existing secrets previously committed must be rotated.
- Created scripts/deploy.sh for reliable local deployments
- Simplified CI pipeline to code quality checks only
- Deploy via: ./scripts/deploy.sh [backend|frontend]
The Gitea Actions runner cannot access host Docker in Coolify
environment, so deployments must be triggered locally.
Changed from Docker-in-Docker to SSH-based deployment:
- Uses rsync to sync code to server
- Builds Docker images on host directly
- More reliable for Coolify environments
- Proper secret management via SSH
- Removed REPO_PATH workaround (use checkout directly)
- Simplified env vars with global definitions
- Fixed network names as env vars
- Updated DATABASE_URL in Gitea secrets
- Cleaner deployment steps
- Better health checks
Problem: We are banned from Afilias (.info/.biz) and Google (.dev/.app)
RDAP servers due to too many requests, causing timeouts.
Solution:
1. Added RDAP_BLOCKED_TLDS list in domain_checker.py
2. Skip RDAP for blocked TLDs, use DNS+WHOIS instead
3. Updated drop_status_checker.py to skip blocked TLDs
4. Removed banned endpoints from RDAP_ENDPOINTS
TLDs now using DNS-only: .info, .biz, .org, .dev, .app, .xyz, .online, .com, .net
TLDs still using RDAP: .ch, .li, .de (working fine)
This prevents bans and timeouts while still providing availability checks.
1. Parallel Zone Downloads (3x faster)
- CZDS zones now download in parallel with semaphore
- Configurable max_concurrent (default: 3)
- Added timing logs for performance monitoring
2. Email Alerts for Ops
- New send_ops_alert() in email service
- Automatic alerts on zone sync failures
- Critical alerts on complete job crashes
- Severity levels: info, warning, error, critical
3. Admin Zone Sync Dashboard
- New "Zone Sync" tab in admin panel
- Real-time status for all TLDs
- Manual sync trigger buttons
- Shows drops today, total drops, last sync time
- Health status indicators (healthy/stale/never)
- API endpoint: GET /admin/zone-sync/status
1. Parallel Zone Downloads (CZDS):
- Downloads up to 3 TLDs concurrently
- Reduced sync time from 3+ min to ~1 min
- Semaphore prevents ICANN rate limits
2. Email Alerts:
- Automatic alerts when sync fails
- Sends to admin email with error details
- Includes success/error summary
3. Admin Zone Sync Dashboard:
- New "Zone Sync" tab in admin panel
- Shows all TLDs with domain counts
- Manual "Sync Now" buttons for Switch/CZDS
- Live stats: drops/24h, total domains
4. Backend Improvements:
- /admin/zone-stats endpoint
- Fixed zone-sync endpoints with correct imports
- Added dnsutils (dig) to backend Dockerfile for DNS zone transfers
- Fixed admin zone sync endpoints with correct imports
- AsyncSessionLocal instead of async_session_maker
Backend:
- Added DROPPING_SOON status to DomainStatus enum
- Added deletion_date field to Domain model
- domain_checker now returns DROPPING_SOON for pending delete
- Track endpoint copies status and deletion_date from drop
Frontend:
- Watchlist shows "TRANSITION" status for dropping_soon domains
- AnalyzePanel shows consistent status from Watchlist
- Status display unified between Drops, Watchlist, and Panel
Backend:
- Fixed datetime timezone error in refresh-all endpoint
- Added _to_naive_utc() helper for PostgreSQL compatibility
Frontend:
- Watchlist now passes domain status to AnalyzePanel
- Status is consistent between Drops, Watchlist, and Sidepanel
- Shows "Available" or "Taken" status in AnalyzePanel from Watchlist
- next.config.js now detects NODE_ENV=production
- Uses http://pounce-backend:8000 in Docker instead of localhost
- Logs backend URL during build for debugging
Backend:
- Fixed track endpoint duplicate key error with proper rollback
- Returns domain_id for already tracked domains
Frontend DropsTab:
- Added trackedDrops state to show "Tracked" status
- Track button shows checkmark when already in watchlist
- Status button shows "In Transition" with countdown
AnalyzePanel:
- Added dropStatus to store for passing drop info
- Shows Drop Status banner with availability
- "Buy Now" button for available domains in panel
- Disabled verify_drops scheduler job (caused RDAP rate limit bans)
- Zone files now saved without RDAP verification (zone diff is reliable)
- Added date-based zone file snapshots with 3-day retention
- Improved DropsTab UI with better status display:
- "In Transition" with countdown timer for dropping_soon
- "Available Now" with Buy button
- "Re-registered" for taken domains
- Track button for dropping_soon domains
- Added --shm-size=8g to backend container for multiprocessing
- Removed duplicate host cron job (scheduler handles everything)
The frontend was calculating wrong time remaining because
end_time was sent without timezone suffix (Z).
JavaScript's new Date() interprets "2025-12-20T20:49:20" as
local time, but the server stores it as UTC.
Fix: Add json_encoders to Pydantic models that append "Z"
to all datetime fields, marking them as UTC.
Affected models:
- AuctionListing
- AuctionSearchResponse
- MarketFeedItem
- MarketFeedResponse
- ScrapeStatus
Zone File Storage:
- Persistent storage in /data/pounce/zones/ (not /tmp)
- 3-day retention for historical snapshots
- Volume mounts in CI/CD pipeline
- New zone_retention.py for snapshot management
Bug Fix:
- Fixed wrong TSIG key for .li zone transfer
- Key was corrupted, causing BADSIG errors
- Now using official Switch.ch key
Config Changes:
- Added switch_data_dir setting
- Added zone_retention_days setting (default: 3)
- CZDS path now defaults to /data/czds
- CI/CD: Add Redis URL and job queue env vars to deploy pipeline
- CI/CD: Fix Frontend BACKEND_URL for internal communication
- Multiprocessing: New zone_file_parser.py with parallel chunk processing
- RAM Drive: Extract zone files to /dev/shm for 50x faster I/O
- CZDS Client: Use high-performance parser with all 32 CPU cores
Performance improvements for Ryzen 9 7950X3D server:
- Zone file parsing: Minutes instead of hours
- Uses ProcessPoolExecutor with 75% of cores
- Memory-efficient streaming for 150M+ domain files
- Single job deployment workflow
- Direct Docker build and deploy on server
- SSL/HTTPS configuration with Let's Encrypt
- Proper Traefik labels for routing
- Health checks and cleanup steps
- Add Gitea Actions workflow for automatic deployment
- Add production Dockerfile for frontend
- Add docker-compose.prod.yml for easy deployment
- Zero-downtime deployment with health checks
- Add StripeService.cancel_subscription() method
- Update /cancel endpoint to call Stripe API
- Set cancelled_at timestamp
- Clean up PostgreSQL reference in server .env
Yield:
- Safely handle partner relationship to prevent errors
Forge:
- AI mode now default for Trader/Tycoon users
- AI mode button moved to left (primary position)
- "Describe your..." label now bright purple
- Textarea border brighter for better visibility
- Check TXT records at ROOT domain instead of _pounce subdomain
- Much simpler for users - just add TXT record with verification code
- Update frontend verification modal with clearer instructions
- Add option to verify immediately when adding domain to portfolio
- Updated yield_dns.py to support A-record verification (simplest method!)
- A-record pointing to 46.235.147.194 is now the primary verification method
- Added Nginx catch-all config for yield domains
- DNS schema updated with method and actual_a fields
- CoreDNS installed but Port 53 blocked by hosting provider
