Multi-Origin Failover Router

Automatic Origin Failover at the Edge — Zero Downtime Routing

A Cloudflare Worker that monitors your primary origin and automatically reroutes traffic to a secondary origin when the primary returns 5xx errors. Configurable retry logic, failure thresholds, and recovery detection — all without touching DNS or your app.

What's Included

• Cloudflare Worker with primary/secondary origin failover logic
• Configurable failure threshold before failover triggers (e.g. 3 consecutive 5xx errors)
• Automatic recovery detection — traffic returns to primary when healthy
• KV-backed failover state shared across all edge locations globally
• Configurable retry count per request before marking origin failed
• Optional alert webhook (Slack, PagerDuty) on failover events
• wrangler.toml with KV binding pre-configured
• README with architecture explanation and tuning guide

How It Works

On each request, the Worker checks a KV flag to determine the current active origin. Requests are forwarded to the primary unless the flag indicates failover mode. When a 5xx response is detected, the Worker increments a failure counter in KV. Once the threshold is hit, the failover flag flips and all subsequent traffic routes to the secondary. A background health check periodically probes the primary and flips the flag back when it recovers.

What You Get

• Transparent origin failover in under 50ms, globally at Cloudflare's edge
• No DNS TTL delays — failover is instantaneous
• KV-backed state ensures all edge nodes fail over simultaneously
• Configurable thresholds prevent flapping on transient errors
• Automatic recovery — no manual intervention needed when primary comes back
• Works with any two HTTP origins: app servers, S3, Render, Railway, etc.

Requirements

• Cloudflare account with Workers and KV enabled
• Two origin URLs (primary and secondary)
• Wrangler CLI installed (npm install -g wrangler)