Consider adding a prometheus /metrics for monitoring kuberos

[SEJeff]

For issues like in #23, it might be good to monitor how many 403s are being received. It might also be nice to see request times to the oidc service along with general health of kuberos. The way dex does it is to include general golang runtime metrics along with some application specific metrics. We use the prometheus operator with a ServiceMonitor to help monitor dex (ultimatey) via grafana dashboards and alerts.

This is a pony feature, but it would be nice!

[negz]

I'd like to implement this, but didn't have time this weekend. I need to Prometheus-ify something during my day job this week so hopefully that will inspire me to find some time to update Kuberos.

If folks have suggestions for metrics please chime in. My thoughts:

• Count of errors, labeled by status code.
• Count of total authentications, labeled by success/failure.
• Total kubecfg template downloads? Not sure if this is useful.

Note Kuberos can't provide details on the calls to the OIDC service because it doesn't actually make any; it just redirects your browser there.

[SEJeff]

Those all seem like valid data points and a great starter. Something > Nothing