Could you please write an examples of Histogram ? about prometheus-cpp HOT 4 CLOSED

There is still a bug open, because bucket boundaries specified through the histogram builder are not used - #46 . A working example can be found at https://github.com/jupp0r/prometheus-cpp/blob/master/tests/family_test.cc#L66

This should work and I'll add it as a test case:

Registry registry{};
auto& histogram_family =
    BuildHistogram().Name("hist").Help("Test Histogram").Register(registry);
auto& histogram = histogram_family.Add({{"name", "test_histogram_1"}},
                                       Histogram::BucketBoundaries{0, 1, 2});
histogram.Observe(1.1);
auto collected = registry.Collect();

from prometheus-cpp.

Hi, do the unit tests at https://github.com/jupp0r/prometheus-cpp/blob/master/tests/histogram_test.cc help answer your question? In general, you use Observe() to count an observation. When a prometheus server is scraping your service, Collect() is called on the histogram to create a snapshot of the values it has counted so far. You usually do not have to call Collect() yourself.

from prometheus-cpp.

I checked the unit test. It seems that it doesn't use the Build* method to create a metric family and call Add() to add a metric the same way as you did in README.md. So may I do like this ?

	auto& hist_cmd_cost_family = BuildHistogram()
			.Name("cerberus_cmd_cost")
			.Help("command cost")
			.Buckets({0,20,30,50}) //is this BucketBoundaries ?
			.Register(*registry);
		auto& hist_cmd_cost = hist_cmd_cost_family.Add({{"cmd_cost","value"}});
		hist_cmd_cost.Observe(proxy->last_cmd_elapse().count());
		hist_cmd_cost.Collect();//just like you said, this is optional

from prometheus-cpp.

Thanks. I'll try that.

from prometheus-cpp.