Command to list caddy servers and their upstreams with health status

[psviderski]

It's a nice idea to list upstreams for served hosts and their health status and latency that should be helpful for troubleshooting. The idea is borrowed from stellar:

[miekg]

wouldn't metrics (also) solve this and be more generic?

cmdline interface is still nice idea as well, but pushing that idea further; the above can then also be shown, but retrieved by asking the in-cluster promtheus server (if we want to have that)

[psviderski]

Maybe if the caddy /metrics endpoint can provide all the required data to list all hosts and their upstreams. Then the rpc call will query /metrics instead of the /reverse_proxy/upstreams endpoint as it's done in #218. Not a big difference tbh.

But I wouldn't artificially add a dependency on the in-cluster Prometheus to serve this data for the CLI if we can easily get them without Prometheus.

I think we shouldn't require every setup to deploy Prometheus in the cluster because it's too heavy. It's also quite often the monitoring stack is deployed separately and shared by multiple clusters/environments so it feels uncloud shouldn't have an opinion here.