Prometheus implementation of MetricRegistry

[tirsen]

I'd like to observe the behavior of the concurrency limits using Prometheus metrics. It's not immediately clear to me how to implement the MetricRegistry using https://github.com/prometheus/client_golang

Could you perhaps provide a quick sketch of how to do that? Or maybe even a full implementation?

[platinummonkey]

this seems relatively straightforward. It just seems to expose an http server with expvar-like query patterns.

could create a new registry that registers the http listener similar to https://github.com/platinummonkey/go-concurrency-limits/blob/master/metric_registry/datadog/registry.go#L50-L119 just swap the dd client reference for the http listener. From here you'd want to register all the available metrics https://github.com/prometheus/client_golang/blob/master/examples/random/main.go#L65-L68 on start and add them during-runtime when they are changed through the Register* interface.

the run loop would be pretty simple: https://github.com/platinummonkey/go-concurrency-limits/blob/master/metric_registry/datadog/registry.go#L143-L148 just swap this out with a pattern used here https://github.com/prometheus/client_golang/blob/master/examples/random/main.go#L83-L84 where it takes the metric samples and "observe"'s them

[jhalterman]

@tirsen Did you ever tackle this? I'm interested in the same thing.