Feature Request - Prometheus/OpenMetrics endpoint

[neufeldtech]

It would be great to have a Prometheus or OpenMetrics endpoint in the HTTP api (/metrics) to be able to gain visibility on how much data is passing through a particular proxy.

A Common question that often comes up when using Toxiproxy in complex environments (with automation, in Kubernetes) is "is my data actually flowing through Toxiproxy?". Having a standard metrics endpoint could help solve that.

Exposing these in a common metric format (Prometheus/OpenMetrics) rather than a custom format (arbitrary JSON or other) would enable wide adoption by folks that already have infrastructure deployed for ingesting Prometheus-style metrics.

For history, see also #266 where metrics were originally proposed (but through a custom format).

[stephenh]

@neufeldtech thank you for this addition!

I'm attempting to use toxiproxy to benchmark database traffic, specifically adding latency & it's working great.

I'd also like to record how many bytes are going over the wire, and these metrics are amazing, with the only wrinkle being that it looks like they're buffered internally, & so not super-amenable to pulling into an automated benchmark script.

I.e. I'm trying to:

• Call /metrics to record startingBytes
• Run my benchmark
• Call /metrics to record endingBytes

But those two numbers are always the same, until ~few seconds/something, and then /metrics shows the updated stats.

Would it be possible to have GET /metrics first ask/force prometheus to flush any buffers it is internally keeping?

Sorry for the late/naive ask--I'll also poke around at this as well. Ty!