hiendt/pgpool repository overview

⁠PostgreSQL High Availability Cluster

Production-ready PostgreSQL 17 HA cluster với automatic failover, connection pooling, và load balancing cho Railway platform.

⁠🐳 Docker Images

⁠PostgreSQL Node (hiendt/postgresql-ha-cluster)

Features:

• PostgreSQL 17 với Repmgr 5.5
• Automatic failover (<10s)
• Streaming replication
• Witness node support
• Optimized for Railway (8GB RAM, 24GB cache)

Usage:

docker run -d \
  --name postgresql-node \
  -e NODE_NAME=pg-1 \
  -e NODE_ID=1 \
  -e IS_WITNESS=false \
  -e PRIMARY_HINT=pg-1 \
  -e POSTGRES_PASSWORD=your_password \
  -e REPMGR_PASSWORD=repmgr_password \
  -e APP_READONLY_PASSWORD=readonly_password \
  -e APP_READWRITE_PASSWORD=readwrite_password \
  hiendt/postgresql-ha-cluster

⁠Pgpool-II (hiendt/pgpool)

Features:

• Pgpool-II v4.3.5
• Connection pooling (max 100 children, 4 per pool)
• Load balancing (primary weight=1)
• Health monitoring
• Watchdog disabled (Railway-specific)

Usage:

docker run -d \
  --name pgpool-node \
  -e POSTGRES_PASSWORD=your_password \
  -e REPMGR_PASSWORD=repmgr_password \
  hiendt/pgpool-ha-cluster

⁠HAProxy (hiendt/haproxy-postgresql-cluster)

Features:

• HAProxy 3.0 Alpine
• Optimized health checks (inter 2s, rise 3, fall 2)
• Automatic recovery from pgpool restarts
• PostgreSQL-aware load balancing

Usage:

docker run -d \
  --name haproxy-lb \
  -p 5432:5432 \
  hiendt/haproxy-postgresql-cluster

⁠🚀 Quick Start - Railway Deployment

Xem hướng dẫn chi tiết trong RAILWAY_DEPLOYMENT.md⁠

⁠Deployment Order

1. Phase 1: Deploy stg-pg-1, stg-pg-2, stg-pg-3, stg-witness (đợi 5-10 phút)
2. Phase 2: Deploy stg-pgpool-1, stg-pgpool-2 (đợi 3-5 phút)
3. Phase 3: Deploy stg-haproxy-1
4. Test: Connect qua HAProxy public URL

⁠Environment Variables

Xem template trong .env.railway - Đổi tất cả passwords!

⁠🏗️ Architecture

7 services: 4 PostgreSQL nodes + 2 Pgpool + 1 HAProxy

HAProxy → Pgpool (2 nodes) → PostgreSQL (1 primary + 2 standbys + 1 witness)

Components:

• PostgreSQL 17: Database engine với streaming replication
• Repmgr 5.5: HA management, automatic failover
• Pgpool-II: Connection pooling & load balancing
• HAProxy: External load balancer
• Witness Node: Quorum cho failover decisions

⁠🔧 Local Development

cp .env.railway .env
# Edit passwords in .env
docker-compose up -d
docker-compose logs -f

⁠📊 Performance Specs

• Memory: 8GB shared_buffers, 24GB effective_cache_size
• Connections: 1000 max per PostgreSQL node
• Failover: <10 seconds
• Load Balancing: Primary reads enabled (weight=1)

⁠🔒 Security

• SCRAM-SHA-256 password encryption
• Application users: app_readonly, app_readwrite
• Admin users: postgres, repmgr
• SSL disabled (Railway internal networking)

⁠📚 Documentation

• RAILWAY_DEPLOYMENT.md⁠ - Full deployment guide
• .env.railway⁠ - Environment variables template
• SCRIPTS_README.md⁠ - Utility scripts

⁠🏷️ Tags

• latest - Latest stable release
• v1.0.0 - Initial production release

⁠🤝 Contributing

1. Fork repository
2. Create feature branch
3. Test locally with docker-compose
4. Submit pull request

⁠📄 License

MIT License - see LICENSE file for details.

Built for Railway | PostgreSQL 17 + Repmgr 5.5 + Pgpool-II v4.3.5 + HAProxy 3.0