I'm using NGINX's ip_hash directive to ensure that client requests always end up at the same backend server. In the case of a backend server outage, NGINX will reroute to the available server, and since I can listen for a disconnection event on my client application, I can have my client-side application react accordingly. However, when the previously down server comes back up, NGINX will route requests back to the original server. This breaks my application, as I have no way of knowing that this happens from my client-side application.

So is there a way for NGINX to send a notification event to my client-side app when it detects that a previously down server is available, or preferably prevent NGINX from exhibiting this behavior?