Crypto infrastructure · Node operations
A blockchain node fleet across 8 networks, with chain-aware health checks
Eight blockchain networks on dedicated servers, built with Terraform and Ansible. A custom health daemon knows when a node has fallen behind the chain and takes it out of rotation on its own.
The situation
Applications that read from or write to blockchains need nodes they can trust. A node that is up but a few blocks behind is worse than a node that is down: it answers quickly, with stale data.
The fleet had to cover 8 networks: Bitcoin, Ethereum, BNB Chain, Solana, XRP, TON, TRON and Litecoin. Each has its own client software, sync behaviour and upgrade cycle.
What we did
Hardware and provisioning.
- Nodes run on dedicated servers, not shared cloud instances, for predictable disk and network performance.
- Servers are provisioned with Terraform and configured with Ansible, so a node can be rebuilt from code.
- All management traffic runs over a mesh VPN.
Health checks that understand the chain.
- A standard health check only asks “is the process running?”. That does not tell you whether the node is in sync.
- We wrote a custom Go daemon that checks each node against the state of its chain.
- HAProxy uses those results to decide where to send traffic. A node that falls behind is taken out of rotation automatically and returns when it has caught up.
Upgrades and monitoring.
- Rolling upgrades are automated, so a network upgrade does not take the whole pool offline at once.
- Prometheus, Grafana and Alertmanager track the fleet, with alerts going to the team.
Outcome
Traffic goes only to nodes that are in sync with their chain. Lagging nodes leave the pool without anyone having to notice them first, and upgrades roll through the fleet one node at a time.
The same setup is what we offer as a service: dedicated blockchain nodes.