Skip to main content
Looking for Contributors! This Skill needs an implementation.Contributing a Skill involves:
  1. Writing a reference implementation with full functionality
  2. Adding comprehensive tests
  3. Writing documentation with usage examples
See the ethers-provider Skill for an example of a complete Skill implementation.Interested? Open an issue or PR at github.com/evmts/voltaire.
Skill — Copyable reference implementation. Use as-is or customize. See Skills Philosophy.
A provider that automatically fails over to backup RPC endpoints when the primary fails. Essential for production reliability.

​
Why Fallback Provider?

Single RPC endpoints fail. Rate limits, downtime, network issues. Production apps need redundancy: 
// Without fallback - single point of failure
const provider = new HttpProvider('https://eth-mainnet.g.alchemy.com/v2/...');
// If Alchemy is down, your app is down

// With fallback - automatic failover
const provider = FallbackProvider([
  'https://eth-mainnet.g.alchemy.com/v2/...',
  'https://mainnet.infura.io/v3/...',
  'https://eth.llamarpc.com',
]);
// Tries each endpoint until one works

​
Planned Implementation

​
Basic Fallback

const provider = FallbackProvider({
  endpoints: [
    { url: 'https://eth-mainnet.g.alchemy.com/v2/KEY', priority: 1 },
    { url: 'https://mainnet.infura.io/v3/KEY', priority: 2 },
    { url: 'https://eth.llamarpc.com', priority: 3 },
  ],
  // Try next endpoint after 5s timeout
  timeout: 5000,
  // Retry failed endpoint after 30s
  retryDelay: 30000,
});

// Uses highest priority available endpoint
const balance = await provider.request({
  method: 'eth_getBalance',
  params: [address, 'latest']
});

​
Quorum Mode

// Require multiple endpoints to agree (for critical operations)
const provider = FallbackProvider({
  endpoints: [...],
  quorum: 2, // Need 2 endpoints to return same result
  quorumWeight: {
    'eth_getBalance': 1,      // Single endpoint OK for reads
    'eth_sendRawTransaction': 2, // Require quorum for sends
  }
});

​
Health Checking

const provider = FallbackProvider({
  endpoints: [...],
  healthCheck: {
    interval: 10000, // Check every 10s
    method: 'eth_blockNumber',
    // Mark unhealthy if >5 blocks behind
    staleness: 5,
  }
});

// Get endpoint health status
const health = provider.getHealth();
// [{ url: '...', healthy: true, latency: 45, blockNumber: 12345678 }, ...]

​
Sticky Sessions

const provider = FallbackProvider({
  endpoints: [...],
  // Stick to working endpoint instead of always trying primary
  sticky: true,
  // Re-check primary every 60s
  primaryCheckInterval: 60000,
});

​
Configuration Options

OptionDescription
endpointsArray of RPC URLs or endpoint configs
timeoutRequest timeout before trying next endpoint
retryDelayWait before retrying failed endpoint
quorumNumber of endpoints that must agree
stickyStick to working endpoint
healthCheckBackground health monitoring config