Topics

#0xfnzero #agave #agave-validator #fnzero #geyser #geyser-plugin #grpc #jito #jito-solana #linux-tuning #nvme #rpc-node #shredstream #solana #solana-node #solana-rpc #solana-validator #ubuntu #yellowstone #yellowstone-grpc

Language breakdown

No language data available.

▸ Files click to expand

README

⚡ Solana RPC Install

Production-ready Solana RPC node deployment in 3 simple steps

Deploy battle-tested Solana RPC nodes with stable, proven configurations and source compilation from GitHub.

中文 | English | Website | Telegram | Discord

What This Project Provides

solana-rpc-install is a focused installer for running a Solana RPC node on Ubuntu with Jito Solana / Agave validator, Yellowstone gRPC, systemd service management, NVMe disk mounting, snapshot bootstrap, and production-oriented Linux tuning.

It is built for operators who need a practical Solana mainnet RPC deployment guide for trading bots, indexing pipelines, DEX event streaming, MEV infrastructure, or private RPC workloads without hand-assembling every validator, Geyser, storage, and monitoring setting.

Included Scripts

| Script | Purpose | |--------|---------| | 1-prepare.sh | Mount NVMe data disks, create Solana directories, and apply Linux system optimizations | | 2-install-jito-validator.sh | Build and install Jito Solana / Agave validator from source | | 3-start.sh | Download snapshots, install the systemd service, and start the RPC node | | validator.sh | Auto-select the right validator profile for 128GB, 192GB, 256GB, or 512GB+ RAM | | yellowstone-config.json | Production-tested Yellowstone gRPC Geyser configuration | | performance-monitor.sh, get_health.sh, catchup.sh | Monitor node health, memory, performance, and sync progress |

🎯 System Requirements

Minimum Configuration:

CPU: AMD Ryzen 9 9950X (or equivalent)
RAM: 128 GB minimum (256 GB recommended)
Storage: 1-3x NVMe SSDs (flexible configuration, script auto-adapts)

- 1 disk: System disk only (basic setup) - 2 disks: System + 1 data disk (recommended, best cost-performance) - 3 disks: System + 2 data disks (optimal performance) - 4+ disks: System + 3 data disks (accounts/ledger/snapshot separated)

OS: Ubuntu 22.04/24.04
Network: High-bandwidth connection (1 Gbps+)

🚀 Quick Start

Three-Step Installation

# Switch to root user
sudo su -
Clone repository to /root
cd /root
git clone https://github.com/0xfnzero/solana-rpc-install.git
cd solana-rpc-install
Step 1: Mount disks + System optimization
bash 1-prepare.sh
(Optional) Verify mount configuration
bash verify-mounts.sh
Step 2: Build Jito Solana from source (15-30 minutes)
bash 2-install-jito-validator.sh
Press Enter to install v4.1.1, or enter a specific version (e.g., v4.1.0-rc.1)
Supports stable, rc, and beta Jito tags
Step 3: Download snapshot and start node
bash 3-start.sh

ℹ️ Installation Method

This installation uses source compilation from GitHub to build Jito Solana validator. This ensures you get the complete agave-validator binary with full MEV support required for RPC nodes.

⚠️ Critical: Memory Management Details (Required for 128GB Systems)

📌 Why Swap Might Be Needed?

- Memory peaks can exceed 128GB during initial sync (115-130GB)

- Without swap, node may crash with OOM

- Swap provides safety buffer during sync phase

- After sync stabilizes, memory usage drops to 85-105GB

🔧 Swap Management (Optional for 128GB Systems)

Add Swap (If needed during sync)

# Only if you see high memory pressure during sync
cd /root/solana-rpc-install
sudo bash add-swap-128g.sh
Script automatically checks:
✓ Only adds swap if system RAM < 160GB
✓ Skips if swap already exists
✓ Adds 32GB swap with swappiness=10 (minimal usage)

Remove Swap (After sync completes)

Once synchronization completes, memory usage stabilizes at 85-105GB, and you can remove swap for optimal performance:

# Check current memory usage
systemctl status sol | grep Memory
If memory peak < 105GB, safe to remove swap
cd /root/solana-rpc-install
sudo bash remove-swap.sh

📊 Decision Guidelines

| Memory Peak | Recommended Action | |-------------|-------------------| | < 105GB | ✅ Can remove swap for optimal performance | | 105-110GB | ⚠️ Recommended to keep swap as buffer | | > 110GB | 🔴 Must keep swap to prevent OOM |

Note: If memory issues occur after removing swap, you can always add it back:

cd /root/solana-rpc-install sudo bash add-swap-128g.sh

🚀 Next Steps: Install Jito ShredStream

After completing your RPC node installation, you can enhance performance with Jito ShredStream:

Quick Start Guide: QUICK_START.md
Repository: jito-shredstream-install

ShredStream provides low-latency block streaming for Jito MEV infrastructure.

📊 Monitoring & Management

# Real-time logs
journalctl -u sol -f
Performance monitoring
bash /root/performance-monitor.sh snapshot
Health check (available after 30 minutes)
/root/get_health.sh
Sync progress
/root/catchup.sh

✨ Key Features

🔧 Battle-Tested Configuration Philosophy

All configurations are based on proven production deployments with thousands of hours of uptime:

Conservative Stability > Aggressive Optimization
Simple Defaults > Complex Customization
Proven Performance > Theoretical Gains

📦 System Optimizations (No Reboot Required)

🌐 TCP Congestion Control: Westwood (classic, stable algorithm)
🔧 TCP Buffers: 12MB (conservative, low-latency optimized)
💾 File Descriptors: 1M limit (sufficient for production)
🛡️ Memory Management: swappiness=30 (balanced approach)
🔄 VM Settings: Conservative dirty ratios for stability

⚡ Yellowstone gRPC Configuration

✅ Compression Enabled: gzip + zstd (reduces memory copy overhead)
📦 Conservative Buffers: 50M snapshot, 200K channel (fast processing)
🎯 Proven Defaults: System-managed Tokio, default HTTP/2 settings
🛡️ Resource Protection: Strict filter limits prevent abuse

🚀 Deployment Features

📦 Source Compilation Installation:

- 🔧 Jito Solana from official GitHub (15-30 min) - ✅ Complete validator binary with full MEV support - 🎯 100% compliant with Jito Foundation standards

🧠 Intelligent Configuration Selection: Auto-detects system RAM and selects optimal validator configuration

- TIER 1 (128GB): Conservative settings for 128-159GB systems - TIER 2 (192GB): Balanced configuration for 192-223GB systems - TIER 3 (256GB): High-performance for 256-383GB systems - TIER 4 (512GB+): Maximum capacity for enterprise deployments

🔄 Automatic Disk Management: Smart disk detection and mounting
🛡️ Production Ready: Systemd service with dynamic memory limits and OOM protection
🌐 Network Resilience: Enhanced version verification with graceful degradation
📊 Monitoring Tools: Performance tracking and health checks included

🔌 Network Ports

| Port | Protocol | Purpose | |------|----------|---------| | 8899 | HTTP | RPC endpoint | | 8900 | WebSocket | Real-time subscriptions | | 10900 | gRPC | High-performance data streaming | | 8000-8030 | TCP/UDP | Validator communication (dynamic) |

📈 Performance Metrics

Snapshot Download: Network-dependent (typically 200MB - 1GB/s)
Memory Usage: 60-110GB during sync, 85-105GB stable (optimized for 128GB systems)
Sync Time: 1-3 hours (from snapshot)
CPU Usage: Multi-core optimized (32+ cores recommended)
Stability: Proven configuration with >99.9% uptime in production

🛠️ Architecture

┌─────────────────────────────────────────────────────────┐
│                   Solana RPC Node Stack                  │
├─────────────────────────────────────────────────────────┤
│  Jito Solana Validator (v4.1.x)                         │
│  ├─ Installation: Source compilation from GitHub        │
│  │  • agave-validator with full MEV support             │
│  │  • 100% Jito Foundation compliant (15-30 min)        │
│  ├─ Yellowstone gRPC auto-matched to Solana version     │
│  ├─ RPC HTTP/WebSocket (Port 8899/8900)                 │
│  └─ Accounts & Ledger (Optimized RocksDB)               │
├─────────────────────────────────────────────────────────┤
│  System Optimizations (Battle-Tested)                   │
│  ├─ TCP: 12MB buffers, Westwood congestion control      │
│  ├─ Memory: swappiness=30, balanced VM settings         │
│  ├─ File Descriptors: 1M limit, sufficient for prod     │
│  └─ Stability: Conservative defaults, proven in prod    │
├─────────────────────────────────────────────────────────┤
│  Yellowstone gRPC (Open-Source Tested Config)           │
│  ├─ Compression: gzip+zstd enabled (fast processing)    │
│  ├─ Buffers: 50M snapshot, 200K channel (low latency)   │
│  ├─ Defaults: System-managed, no over-optimization      │
│  └─ Protection: Strict filters, resource limits         │
├─────────────────────────────────────────────────────────┤
│  Infrastructure                                          │
│  ├─ Systemd Service (Auto-restart, graceful shutdown)   │
│  ├─ Multi-disk Setup (System/Accounts/Ledger)           │
│  └─ Monitoring Tools (Performance/Health/Catchup)       │
└─────────────────────────────────────────────────────────┘

🧪 Configuration Philosophy

Why Conservative Configuration?

Based on extensive production testing, we discovered:

Compression Enabled = Lower Latency

- Even on localhost, compressed data transfers faster in memory - CPU overhead is minimal, latency reduction is significant

Smaller Buffers = Faster Processing

- 50M snapshot vs 250M: Less queue delay, faster throughput - 200K channel vs 1.5M: Reduced "buffer bloat" latency

System Defaults = Better Stability

- No custom Tokio threads: Let system auto-manage - No custom HTTP/2 settings: Defaults are already optimized - Fewer custom parameters = Fewer potential issues