One primary origin plus ordered backups, each with its own health check
Every domain has a primary origin — the one marked default — and zero or more backups ordered by priority. Traffic uses the primary until health checks or the circuit breaker say otherwise, at which point a healthy backup takes over without publishing a new configuration.
1. Open the Origins section#
On the domain detail page choose the Setup category, then the Origins section (described in the menu as Backend source servers and health checks.). This is the same place the onboarding flow's Add origin button writes to, so origins added during setup appear here too.
2. Add a backup origin#
Add the origin with its URL, the health path to probe, its timeouts, and an optional Host-header override if the backend expects a specific hostname. Each origin carries its own health-check settings, so a backup can be checked more aggressively than the primary.
3. Configure health checks#
Each origin supports GET or HEAD probes with a path, expected status codes, an interval, a timeout, and separate healthy and unhealthy thresholds — separate thresholds mean a single blip never flaps the origin. Only idempotent GET, HEAD, and OPTIONS requests are ever retried or failed over; writes make exactly one attempt, so a request with an unknown outcome is never replayed.
4. Change which origin is default#
Promote another origin to move the default badge. Non-default origins can be deleted; the default cannot, because removing it would leave the domain with nowhere to send traffic. A manual check updates an origin's health and its checked-at timestamp on demand, which is the fastest way to confirm a fix worked.
Expected result#
The Origins list shows one default origin plus any backups, each with its current health state. Failover happens inside the proxy, so no configuration change and no deploy is needed when an origin is marked unhealthy.