RPC Failover

Overview

What it means and why it exists

RPC failover is the process of switching blockchain data requests or transaction submissions to another node or provider when the primary endpoint is unavailable or unreliable.

How it works

A service monitors health, chain ID, block height, latency and error rates across configured RPC endpoints. When the primary fails, it routes safe queries to another endpoint and may rebroadcast already signed transactions as appropriate. It preserves transaction references and compares responses to detect stale or inconsistent chain views.

Technical details worth understanding

For reads, failover generally improves availability but does not remove the need to assess finality. For writes, resending the same signed raw transaction is different from creating a new transfer; the former normally keeps the same transaction identity, while the latter can risk duplication. A good design distinguishes transport-level retries from business-level retries.

What can go wrong

Two endpoints can both be live but disagree temporarily because of chain reorganizations, indexing lag or configuration differences. A fallback node on the wrong network can produce misleading results. Shared cloud infrastructure can also create a hidden single point of failure.

Practical checks

Test that secondary RPCs use the same chain ID and are caught up to the network. If a signed transaction must be rebroadcast, send the same raw transaction rather than constructing a new beneficiary payment. Monitor when providers disagree about block inclusion or logs, and hold irreversible fulfillment until the discrepancy is resolved.

Related concepts

Payment Status API, Webhook, API Key.

Further reading

OWASP — REST Security Cheat Sheet.