Proposal

As we have grown our use of KEDA, we would like to be able to better observe our outbound requests to our Prometheus Instance to understand and observe this very critical code path. At the moment, all we have are the internal scaler error metrics, but we would like to observe the Prometheus specific response codes and have the metrics to be able to compute a success rate metric for this request path

Use-Case

• Be able to observe prometheus specific api query errors and seperate them from the Scaler Errors metric
• Compute a success rate metric that we can internally alert on to give us warning of a prometheus issue

Is this a feature you are interested in implementing yourself?

Yes

Anything else?

Happy to implement this. The current GRPC and Scaler metrics today are lacking detail on prometheus specific errors so we are unable to discern, for example, the difference between the upstream endpoint being down vs being rate limited.