We have NGINX+ R24 running as API proxy. We implement the max_conns parameter for upstream servers, ie.

upstream <cluster> {
    zone  <cluster> 512k;
    server <server1-ip>:<port> max_conns=32;
    server <server2-ip:<port> max_conns=32 ;

We have a grafana dashboard for the metric nginxplus_upstream_server_active but additionally want a metric counting how often a request could not be proxied to an upstream server because the max_conns limit was already hit for that server.

When looking at the NGINX+ API, it does not seem that counter is available so that would have to be implemented as well.