Skip to main content

Multiple origins & failover

On Pro and Business plans, a property can have multiple origins with a priority order. The edge always fetches from the highest-priority enabled origin; if that origin fails, it automatically retries the next one — so a dying backend doesn't take your site down. See Plans & pricing.

How failover works​

Origins are tried in priority order (1 first). The edge moves on to the next enabled origin when the current one:

  • Fails to connect — connection refused, connect/read timeout, or a TLS failure.
  • Returns a 5xx response — 500, 502, 503, 504, and friends.

A 4xx response does not trigger failover — a 404 or 403 is a valid answer from a healthy origin and is served as-is.

The retry happens within the same request — the visitor just gets the response from the first origin that answers, a little slower on the failed attempt. When the primary origin recovers, traffic returns to it automatically; there is nothing to reset.

Adding origins​

  1. Open your property and go to the General tab.
  2. In the Origin Server card, click Add origin and enter the host — the same forms as the first origin are accepted, see Allowed origins.
  3. Set the origin's priority — 1 is tried first. Reorder origins at any time; the order is the failover order.
  4. Each origin has its own Host Header override and SSL verify toggle, and can be enabled or disabled individually — a disabled origin is skipped entirely, which is handy for taking a backend out of rotation during maintenance.
  5. Save the card.
Origins must serve the same content

Failover assumes every origin can answer any request for the property. Point the origins at replicas of the same site — not at different applications — or visitors will see inconsistent content whenever a failover occurs.

With Origin Shield

If the property has an Origin Shield enabled, failover happens at the tier that contacts your origins: the shield walks the priority list, and the edge keeps talking to the shield as usual.