Add support for pushint oltp metrics

[tiny-pangolin]

Issue Details

I would like the ability to have metrics pushed in otel format so I can resuse dashboards and alerting rules between caddy and other reverse proxies that follow otel semantic conventions

Assistance Disclosure

AI not used

If AI was used, describe the extent to which it was used.

No response

[francislavoie]

That's already supported https://caddyserver.com/docs/metrics#caddys-metrics unless I misunderstand

[tiny-pangolin]

Those metrics are exposed in Prometheus format which is what I'm using now, but it would nice if caddy could be configured to push the metrics using the otlp protocol and follow the semantic conventions for http requests

[francislavoie]

Okay, seems related to #5743. Duplicate?

As I wrote there, most of Caddy's core maintainers don't use tracing or metrics ourselves, so we need to lean on the community to contribute improvements.

[trajano]

I don't think #5743 would handle it yet, metrics are tackled differently than tracing.

Like you @tiny-pangolin I use the prometheus endpoint from otel-collector to scrape my caddy server (named edge)

  27   │   prometheus:
  28   │     config:
  29   │       scrape_configs:
  30   │         - job_name: caddy
  31   │           static_configs:
  32   │             - targets: ['edge:2019']

I get the metrics on grafana which I export using otlphttp from otelcollector.