Full Ethereum Node Deployment: Geth and Lighthouse for Secure RPC Access
Launching an Ethereum node is easier now than before, but there are still pitfalls. The main one: after The Merge, you need to run two clients simultaneously: Execution Layer (EL) and Consensus Layer (CL). They communicate via Engine API with JWT authentication, and without CL, the node simply won't sync. The complexity lies in incorrect configuration of one of the clients or ignoring JWT, leading to engine_exchangeTransitionConfiguration errors. Our team of 20 blockchain engineers has deployed over 50 full nodes in production since 2016, so we know all the typical mistakes. We guarantee 99.9% uptime with our monitoring setup.
Choosing the Right Client Combination
Execution Layer: Geth is the reference client in Go, with ~70% market share. Snap sync completes the initial state download in 4–12 hours, leveraging the historical state snapshot. Nethermind (.NET) is good for Windows servers. Erigon (Go) offers the smallest database size (~1 TB vs 1.2 TB for Geth), but snapshot sync is slower. Besu (Java) is suitable for enterprise.
Consensus Layer: Lighthouse (Rust) delivers excellent performance and low memory consumption. The Beacon Chain client processes attestations and block proposals, requiring low-latency connections. Prysm (Go) is the most popular with good documentation. Teku (Java) is an option for large organizations.
Recommended combination: Geth + Lighthouse — both are stable and well-documented. Geth snap sync is 40% faster than Erigon, and Lighthouse consumes 30% less memory than Prysm.
| Client (EL) | Database Size | Snap Sync Speed | Memory (RAM) | Features |
|---|---|---|---|---|
| Geth | 1.2 TB | 4–12 h | 4–6 GB | Reference, stable, largest community |
| Erigon | ~1 TB | 12–24 h | 4–6 GB | Minimal DB size, slower sync |
| Nethermind | 1.2 TB | 6–14 h | 4–6 GB | .NET, good for Windows |
| Besu | 1.5 TB | 8–16 h | 6–8 GB | Java, enterprise features |
Hardware Requirements
| Node Type | Disk | RAM | CPU |
|---|---|---|---|
| Full node (snap sync) | 1.2 TB NVMe | 16 GB | 4 cores |
| Archive node | 14+ TB NVMe | 32 GB | 8 cores |
| Validator | 1.2 TB NVMe | 16 GB | 4 cores |
Critical: Only NVMe SSDs. HDD and SATA SSDs won't provide enough IOPS — the node will lag behind.
Why JWT Secret Is Critical
After The Merge, communication between Execution and Consensus clients is secured with JWT (JSON Web Token). Without this file, the Execution and Consensus clients cannot authenticate inter-process communication via the Engine API, resulting in a failure to synchronize. Generation is simple:
openssl rand -hex 32 > jwt.hex chmod 600 jwt.hex Ensure the paths to jwt.hex in both clients are configured correctly. An error in --authrpc.jwtsecret or --execution-jwt is one of the most common deployment mistakes.
How to Ensure Smooth Synchronization
Use --checkpoint-sync-url for the Consensus Layer. This speeds up CL sync from weeks to 15 minutes by downloading the finalized beacon state. Public endpoints: https://mainnet.checkpoint.sigp.io, https://beaconstate.ethstaker.cc. For the Execution Layer, use snap sync — Geth downloads only the state snapshot in 4–12 hours. The combination of checkpoint sync + snap sync gives maximum speed.
Deployment Steps
- Choose your client combination (recommended: Geth + Lighthouse).
- Provision hardware with NVMe SSD, 16 GB RAM, 4+ CPU cores.
- Generate JWT secret and configure both clients.
- Set up Docker Compose with proper network and volume mappings.
- Start clients and monitor sync progress via eth_syncing and beacon APIs.
Monitoring
curl -s -X POST http://localhost:8545 -H "Content-Type: application/json" -d '{"jsonrpc":"2.0","method":"eth_syncing","params":[],"id":1}' | jq curl -s http://localhost:5052/eth/v1/node/syncing | jq Add to Prometheus via ethereum_exporter and lighthouse_beacon. Dashboards from official monitoring repos allow you to track latency, peer count, and sync status.
RPC Security
- HTTP RPC (8545, 8546) must not be exposed to the internet without authentication — anyone could drain unlocked accounts or overload the node with
eth_callattacks. - Use Nginx reverse proxy with rate limiting and, if external access is needed, Bearer token authentication.
- Engine API (8551) — only localhost or internal Docker network.
- Firewall: only P2P ports open (30303, 9000).
Use iptables or cloud firewall to block incoming connections to 8545 from outside.
What's Included in Turnkey Node Deployment
We provide: client selection and configuration, JWT secret generation, Docker Compose setup, monitoring configuration (Prometheus + Grafana), security hardening (firewall, Nginx reverse proxy), configuration documentation, team training, and one month of technical support. Our turnkey deployment starts at $1,500 for a full node setup, with ongoing support at $200/month.
How We Work
Our process: gather requirements → audit current infrastructure (if any) → design node architecture → provide a fixed-price estimate → deploy and test → handover with documentation. Pricing is determined after analysis, typically within 2 working days. We have been in the blockchain infrastructure market for 5 years, and our engineers hold certifications in Ethereum development and DevOps.
Typical Mistakes (Checklist)
Common errors when deploying
- Using HDD or SATA SSD — node falls behind.
- Forgetting JWT secret — clients can't connect.
- Exposing HTTP RPC without authentication — attack risk.
- Syncing from genesis instead of checkpoint/snap — weeks of waiting.
- Misconfiguring Engine API port (8551) —
engine_exchangeTransitionConfigurationerror.
Case Study: DeFi Platform Node Deployment
For one of our clients, a DeFi platform requiring low-latency RPC access, we deployed a Geth + Lighthouse node. By using snap sync and checkpoint sync, we reduced initial sync time from 3 days to 6 hours. Post-deployment, RPC response times improved by 40% compared to their previous third-party provider, and we set up Grafana dashboards for real-time monitoring. The node has been running for 8 months with 99.9% uptime, as guaranteed.
Contact us for a detailed consultation. Order a turnkey node deployment starting from $1,500 — we'll evaluate your project within 2 working days. Save up to 30% compared to cloud providers.