We incorrectly didn't use the returned `Responder` if the client had disconnected, which meant that the resource used by the Responder wasn't correctly released. In particular, this exhausted the thread pools so that *all* requests timed out.
Erik Johnston