Please add a metric to count when it is hit and the return status.

would a metric be useful here? I dont think people want to set an alert for a metric to know when a debugging endpoint was requested?

@dashpole or @dgrisonnet for thoughts about whether we need metrics for zpages.

@johnbelamaric after thinking more on this, I think we could delegate metrics for these endpoints for future only if/when the need arises to monitor these endpoints. Generally we want these endpoints to behave the same way as other existing debugging/monitoring endpoints like /healthz, /livez, /metrics - we dont have dedicated metrics for these endpoints too.

Does that sound ok to you?

I think each component should instrument the /flagz and /statusz endpoint similarly to the health endpoints today. For example the apiserver records requests made to these endpoints in its apiserver_request_total and apiserver_request_duration_seconds metrics: https://github.com/kubernetes/kubernetes/blob/master/staging/src/k8s.io/apiserver/pkg/server/healthz/healthz.go#L212

Although the zpages endpoints are unlikely to be DDOSed, there is always a possibility and we would need this information in the http metrics to be aware of that situation

Yeah. Abuse is the main thing it can help identify.

Oh I missed noting that we do have coverage for healthz, livez, readyz endpoints via the apiserver_request_total and apiserver_request_duration_seconds metrics. Thanks for pointing that out!

Updated the KEP to make flagz follow the same pattern. Also updated statusz KEP to state the same here #5910

Thanks both for the discussion!