Document updated on May 15, 2026

Fallback Proxy

The Fallback Proxy enables performing an alternative backend request when the previous one has failed. Although calling a fallback will increase the latency, there are some cases where it is preferable to call a different service so the request isn’t lost instead of returning a failure that the client would need to handle.

Fallback proxy configuration

To enable the fallback proxy, you need to add in the endpoint definition the following configuration:

{
    "endpoint": "/hotels/{id}",
    "extra_config": {
          "proxy": {
              "strategy": "fallback"
          }
      }
}

After doing this, the backend list is treated as a prioritized list, executed one by one, where the next backend will only be executed if the current one fails.

For example imagine you have a backend that handles your register flow, but instead of failing when that backend fails for whatever reason, you want to send the otherwise failed requests to a queue to be processed in the background by using AMQP Producer component.

{
  "endpoint": "/register",
  "extra_config": {
    "proxy": {
        "strategy": "fallback"
    }
  },
  "backend": [
    {
        "url_pattern": "/__echo/register",
        "method": "POST",
        "host": [ "http://unknown" ]
    },
    {
      "host": ["amqp://guest:[email protected]:5672"],
      "disable_host_sanitize": true,
      "extra_config": {
          "backend/amqp/producer": {
              "exchange": "some-exchange",
              "durable": true,
              "delete": false,
              "no_wait": true,
              "no_local": false,
              "routing_key": "register.error",
              "mandatory": false,
              "immediate": false,
              "backoff_strategy": "exponential-jitter"
          }
      }
    }
  ]
}