Prometheus support for go-metrics

Definitely something stupid I'm doing but can't figure it out. Prometheus shows all the default ones, but not the metrics from the go-metrics examples:

prometheusRegistry := prometheus.NewRegistry()
prometheusClient := prometheusmetrics.NewPrometheusProvider(metrics.DefaultRegistry, "", "", prometheusRegistry, 1*time.Second)
go prometheusClient.UpdatePrometheusMetrics()

http.Handle("/metrics", promhttp.Handler())
log.Println("Beginning to serve on port :8080")
log.Fatal(http.ListenAndServe("127.1:8080", nil))

If I create promauto metrics they appear. Any hint?

This change fixes the module name in the go.mod file following PR #11, which resulted in broken importing:

$ go get ...
...
go get: github.com/deathowl/[email protected] updating to
	github.com/deathowl/[email protected]: parsing go.mod:
	module declares its path as: github.com/tumani1/go-metrics-prometheus
	        but was required as: github.com/deathowl/go-metrics-prometheus

Use the flattened name for the help argument when building the prometheus description since the name parameter uses flattened version as well.

Without this, you can get an error like: has help "name_for-example" but should have "name_for_example" (notice the difference between the hyphen and the underscore between for and example)

Need a quick fix here so that we don't run into the issue for go get

go get: github.com/deathowl/[email protected]: parsing go.mod:
        module declares its path as: github.com/tumani1/go-metrics-prometheus
                but was required as: github.com/deathowl/go-metrics-prometheus

Replication: Start application with —race flag, Try to get metrics several times. Frequently we catch race condition.

Trace log: Previous write at 0x00c00019c7f0 by goroutine 97: github.com/prometheus/client_golang/prometheus.NewDesc() /Users/i.tumanov/go/pkg/mod/github.com/prometheus/[email protected]/prometheus/desc.go:81 +0x8d github.com/deathowl/go-metrics-prometheus.(*PrometheusConfig).histogramFromNameAndMetric() /Users/i.tumanov/go/pkg/mod/github.com/deathowl/[email protected]/prometheusmetrics.go:118 +0xbda github.com/deathowl/go-metrics-prometheus.(*PrometheusConfig).UpdatePrometheusMetricsOnce.func1() /Users/i.tumanov/go/pkg/mod/github.com/deathowl/[email protected]/prometheusmetrics.go:163 +0x72a github.com/rcrowley/go-metrics.(*StandardRegistry).Each() /Users/i.tumanov/go/pkg/mod/github.com/rcrowley/[email protected]/registry.go:68 +0x1cc github.com/deathowl/go-metrics-prometheus.(*PrometheusConfig).UpdatePrometheusMetricsOnce() /Users/i.tumanov/go/pkg/mod/github.com/deathowl/[email protected]/prometheusmetrics.go:148 +0xfb github.com/deathowl/go-metrics-prometheus.(*PrometheusConfig).UpdatePrometheusMetrics() /Users/i.tumanov/go/pkg/mod/github.com/deathowl/[email protected]/prometheusmetrics.go:143 +0xe8

default bucket values on instantiation with option to globally set custom buckets integration test for histogram metrics

At least partially resolves #9

prometheus officially requires golang 1.5. This patch is removing 1.4 from the travis build configuration (as it is failing currently) in favor of adding golang 1.7 and 1.8 build steps.

Since the project is not private, and it's been linked to from the original go-metrics project I'm guessing the project is intented to be Open Source.

I'd like to make some additions and submit them for review, however clarifying that the project is open source would be a healthy formality. Addresses #2

I've created a LICENSE file with copyright given to: Copyright 2016 Csergő Bálint github.com/deathowl

Please consider if you'd like contributions. Cheers!

Hi hi,

I'm evaluating various monitoring systems and libraries to use with a Go project. This repo came up and looks promising. One thing that would help me is a clarification of what open source license this code is available under.

Thanks!

Hi.

Meters and Timers have quite a lot of methods to acquire rates, counts and other stuff, but not all of them are published to Prometheus registry. I suppose they can be useful in certain situations, thus this PR adds missing metrics to registry as gauges.

Some of the metrics are still missing (like percentiles) - my guess is to choose a number of the most useful of them (for example, 25%, 50%, 75%, 90%, 95%, 99%) and add them as gauges as well.

Big thanks for the library, by the way :)

If the target Prometheus Registry has some metrics already registered (for instance, by an previous run of UpdatePrometheusMetrics()), trying to create a new (*PrometheusConfig) targetting the same Prometheus Registry will produce a panic, since MustRegister() will die if an error occurs.

Simply replacing MustRegister() with Register() will enable this flow without killing the process, since MustRegister() will not panic, but instead return an error.

Why is this an issue? In our use case, we instantiate a server which creates the goroutine with the pClient.UpdatePrometheusMetricsOnce(). When running in production-ish mode, this happens exactly once per process. But when running the tests, we often create and destroy this server several times in a row, in the same process, along with the goroutine that calls pClient.UpdatePrometheusMetricsOnce(). As it goes, the second time that happens, the call to MustRegister() within gaugeFromNameAndValue() abruptly ends the process with a nice panic, which kinda spoils the fun for us.

We've changed MustRegister() with the more lax Register(), aiming towards the idempotence of the update process.

Note that in no way are we trying to run two updaters concurrently.

Thanks for your time.

Co-authored-by: Jose Luis Lucas [email protected]

https://github.com/deathowl/go-metrics-prometheus/blob/master/prometheusmetrics.go#L103 snapshot.Percentiles(buckets) returns a slice of arbitrary percentiles of values in the sample

It seems that it should correspond to the Summary in Promethus, right ? why use histogram ?

I'm trying to collect metrics of my service into prometheus using this rough implementation:

pc := prometheusmetrics.NewPrometheusProvider(someRegistry, someNameSpace, someSubSystem, prometheus.DefaultRegisterer, time.Second)
go pc.UpdatePrometheusMetrics()

http.Handle("/metrics", promhttp.Handler())
err := http.ListenAndServe(":8080", nil)
if err != nil {
	// stop UpdatePrometheusMetrics
}

Is there any way to stop UpdatePrometheusMetrics()? I don't want to leak an unnecessary goroutine if the endpoint is not even accessible.