Support max_fails and keepalive configuration for upstreams

SafeLine 9.3.11 generates every HTTP upstream with hardcoded:

keepalive 128;
keepalive_timeout 75;

and does not expose max_fails/fail_timeout in the UI or Open API.

With multiple Kubernetes NodePort upstreams, rolling updates can produce failed
or stale connections. Because max_fails defaults to 1, all peers can temporarily
become unavailable and Tengine returns "no live upstreams".

Manual changes to IF_backend_<site_id> are not persistent because reloadWebsite
and reloadFullWebsite regenerate these files. Is there a supported site-level
way in 9.3.11 to configure:

1. server max_fails and fail_timeout;
2. upstream keepalive and keepalive_timeout;
3. arbitrary directives inside the generated upstream block?

If not, is this planned, and is there an issue/roadmap item we can follow?

Currently, custom configuration is only supported inside the location / block. Customizing the upstream {} block isn’t supported yet.
I’ve submitted an issue to our product and R&D team for your request, so they can evaluate adding support for this.