Adaptive routing controls features that modify the routing of requests to pools and endpoints in response to dynamic conditions, such as during the interval between active health monitoring requests.
Zero-downtime failover will trigger a single retry only if there is another healthy endpoint in the pool and a 521, 522, 523, 525 or 526 error code is occurring. No other error codes will trigger a zero-downtime failover operation.
When there are no healthy endpoints in the same pool, failover across pools extend the zero-downtime failover of requests to healthy endpoints in alternate pools according to the failover order defined by traffic and endpoint steering.
-
In the Cloudflare dashboard, go to the Load Balancing page.
Go to Load Balancing ↗ -
Navigate to your Load Balancers and select Edit.
-
From Adaptive Routing, enable Failover across pools.
When an origin sends a GOAWAY frame, Cloudflare stops sending new requests on that connection but does not mark the endpoint as unhealthy. Safe-to-retry requests (typically GET) are automatically retried on a new connection. Non-idempotent requests (such as POST or PUT) may not be retried unless the request was not yet sent on the closing connection.