Solana Node Deployment: Setup and RPC Access Optimization

Why Running a Solana Node Is Non-Trivial

In our practice, we encountered situations where a client purchased a server meeting the official minimum requirements, but the node fell behind the tip within a week. Solana is one of the most resource-demanding blockchains: minimum system requirements are not approximate recommendations but a hard threshold below which the node simply cannot keep up with the network and degrades. Wrong disk choice, insufficient RAM, or saving on the network channel leads to constant forks and loss of RPC stability. We have configured nodes for DeFi projects processing thousands of requests per second and developed a configuration that keeps slot lag below 10 slots. With over 5 years of experience in Solana infrastructure and 50+ deployed nodes, we guarantee reliable operation. Our optimized configuration can save up to 30% on hardware costs compared to trial-and-error approaches, translating to annual savings of $3,000–$5,000 for a typical RPC node. Contact us to select the optimal hardware for your load.

Hardware Requirements: What Really Matters

The official Solana Foundation requirements (Solana docs) and the actual production minimum diverge. Practical numbers as of now:

RAM Requirements: Solana stores account state in memory (accounts DB). With over 1.8 billion accounts on the network, this takes hundreds of gigabytes. A node with 128 GB RAM will not run stably.

NVMe Requirements: I/O speed is critical. The node processes thousands of transactions per second, writes the ledger, and responds to RPC—all simultaneously. Rotational HDDs for accounts/ledger are unacceptable: they are 50 times slower in random access throughput than NVMe. NVMe with PCIe 4.0 is 2x faster than PCIe 3.0 for Solana workloads.

Choosing an Optimal Configuration for an RPC Node

If you need an RPC node without voting, you can save on CPU (12 cores is enough), but not on RAM and disks. For a commercial RPC serving tens of thousands of requests, take 512 GB RAM and 4 TB NVMe with PCIe 4.0. Network – definitely 10 Gbps, otherwise clients will complain about timeouts. The right configuration pays off due to the absence of downtime—a single minute of downtime can cost $1,000 for a DeFi project.

Installation and Configuration

Follow these steps to deploy your node.

System Preparation

# Ubuntu 22.04 LTS — recommended OS # Sysctl tuning for high load cat >> /etc/sysctl.conf << EOF net.core.rmem_max = 134217728 net.core.wmem_max = 134217728 net.core.rmem_default = 134217728 net.core.wmem_default = 134217728 net.core.optmem_max = 134217728 net.core.netdev_max_backlog = 65536 EOF sysctl -p # Limits for solana-validator process cat >> /etc/security/limits.conf << EOF solana soft nofile 1000000 solana hard nofile 1000000 solana soft memlock unlimited solana hard memlock unlimited EOF # Hugepages for memory performance vm.nr_hugepages = 131072 >> /etc/sysctl.conf 

Solana CLI Installation

# Install specific version — not latest in production SOLANA_VERSION="v1.18.26" sh -c "$(curl -sSfL https://release.solana.com/${SOLANA_VERSION}/install)" export PATH="/home/solana/.local/share/solana/install/active_release/bin:$PATH" solana --version 

RPC Node Configuration

# /home/solana/start-validator.sh #!/bin/bash exec solana-validator \ --identity /home/solana/validator-keypair.json \ --known-validator 7Np41oeYqpe1GAUzqNoFdJ5SAAQhphFp8s6XAXFCLRiE \ --known-validator GdnSyH3YtwcxFvQrVVJMm1JhTS4QVX7MFsX56uJLUfiZ \ --known-validator DE1bawNcRJB9rVm3buyMVDbezCfKkKa3aTEnDqeS89UB \ --only-known-rpc \ --rpc-port 8899 \ --private-rpc \ --dynamic-port-range 8000-8020 \ --entrypoint mainnet-beta.solana.com:8001 \ --entrypoint entrypoint2.mainnet.solana.com:8001 \ --entrypoint entrypoint3.mainnet.solana.com:8001 \ --expected-genesis-hash 5eykt4UsFv8P8NJdTREpY1vzqKqZKvdpKuc147dw2N9d \ --wal-recovery-mode skip_any_corrupted_record \ --ledger /mnt/ledger \ --accounts /mnt/accounts \ --snapshots /mnt/snapshots \ --log /home/solana/solana-validator.log \ --limit-ledger-size 50000000 \ --no-voting \ --enable-rpc-transaction-history \ --enable-extended-tx-metadata-storage \ --rpc-bind-address 0.0.0.0 

Key flags:

• --no-voting — RPC node, not a validator (no stake needed)
• --limit-ledger-size 50000000 — limits ledger size (~200 GB). Without this, the ledger grows indefinitely
• --enable-rpc-transaction-history — store transaction history (needed for getTransaction)
• --known-validator — trusted validators for initial sync. Mandatory for security, otherwise the node may sync to a fork

First Launch – Snapshot Sync

Syncing from genesis takes weeks. Use a snapshot — it is 10x faster:

# Download latest snapshot from official sources # List of available snapshots: https://api.mainnet-beta.solana.com/ solana-validator \ --ledger /mnt/ledger \ download-latest-snapshot \ --snapshot-dir /mnt/snapshots \ --trusted-validators 7Np41oeYqpe1GAUzqNoFdJ5SAAQhphFp8s6XAXFCLRiE 

After downloading the snapshot (~100+ GB), the node starts and catches up to the tip in a few hours.

Node Monitoring

# Node status solana-validator --ledger /mnt/ledger monitor # Health information curl -s http://localhost:8899 -X POST -H "Content-Type: application/json" \ -d '{"jsonrpc":"2.0","id":1,"method":"getHealth"}' | jq . # Lag behind tip (should be < 100 slots normally) solana catchup --our-localhost 8899 

Critical metrics: slot lag, skipped slots, memory usage (should be < 90% RAM), I/O wait. Common metrics include leader schedule slots, turbine retransmissions, and gossip reachability.

Prometheus + Grafana: the solana-exporter publishes metrics in Prometheus format. Dashboards are available on Grafana Marketplace.

Key Metrics Indicating Problems

If slot lag exceeds 1000 slots and skipped slots increase, it signals that the node cannot handle the load. Look at I/O wait: if > 30%, the disk is a bottleneck. When RAM is fully loaded with the accounts database – monitor approaching 90%. Configure alerts in Grafana for these thresholds.

Typical Problems

Node falls behind and cannot catch up: usually an I/O issue – the accounts database cannot keep up. Check iostat -x 1, if await > 50 ms, you need a faster NVMe.

OOM killer kills the process: 256 GB RAM is at the limit. Solution: add swap on NVMe (not HDD), set vm.swappiness=10.

Node forks: check --known-validator and --expected-genesis-hash. A node without trusted validators is vulnerable to eclipse attacks.

How We Configure a Solana Node: Work Stages

We offer turnkey Solana node deployment, from requirement analysis to handover.

Timelines are approximate: 4 to 8 working days. Cost: from $5,000 for a basic RPC node setup (excluding hardware). Contact us for a project assessment.

Deliverables and What's Included

• Server preparation: sysctl, limits, hugepages, disk partitioning
• Installation and configuration of solana-validator
• Initial sync via snapshot
• Systemd service and auto-restart setup
• Monitoring: Prometheus metrics, Grafana dashboard, alerts on slot lag
• Documentation: detailed configuration and maintenance guide
• Access: SSH and admin credentials
• Training: 2-hour session for your team
• Support: 30 days post-launch

Order Solana node setup — we ensure stable RPC access with minimal downtime. Get a consultation from an engineer.