Easier way to specify histogram boundaries?

[nwalters512]

NB: Before opening a feature request against this repo, consider whether the feature should/could be implemented in the other OpenTelemetry client libraries. If so, please open an issue on opentelemetry-specification first.

Is your feature request related to a problem? Please describe.

It's currently somewhat unergonomic to create a histogram with custom buckets/boundaries. Because this must be done with Views, and because Views can only be provided when a MeterProvider is created, this means that the boundary configuration for a histogram must live in a completely separate part of the codebase from the createHistogram call itself. In most cases I can imagine, the buckets for a given histogram are going to be tightly coupled to what individual histograms are measuring. For example, the boundaries for a histogram tracking SQL query duration are likely to be very different from the boundaries for a histogram tracking average response size, and in both cases I'd like to specify those boundaries when I'm creating the histogram metric.

Describe the solution you'd like

I'd like to be able to define a histogram's boundaries when I call createHistogram.

Per #2999, it looks like this used to be supported by the API, though I understand that this was removed in favor of using a View.

If a View is the way to go, perhaps it would be possible to provide a View in the call to createHistogram? Maybe something like this:

meter.createHistogram('foo.duration', {
  view: new View({
    aggregation: new ExplicitBucketHistogramAggregation([0, 1, 5, 10, 15, 20, 25, 30]),
  }),
});

Alternatively, one could directly provide an aggregation and a View would be transparently create for the metric:

meter.createHistogram('foo.duration', {
  aggregation: new ExplicitBucketHistogramAggregation([0, 1, 5, 10, 15, 20, 25, 30]),
});

Describe alternatives you've considered

AFAICT, there aren't any currently viable alternatives to creating a View alongside a MeterProvider

Additional context

This feature would make it possible to easily share and reuse histogram metrics via libraries. This would be really useful for e.g. building a library that can capture common Node metrics, including a histogram for GC duration with bucket sizes that have a sensible default. I'm thinking about something akin to this from the Prometheus world: https:/siimon/prom-client/blob/8d6456c05f0708fed34628da5bd6f492a0ab43f8/lib/metrics/gc.js.

[weyert]

You might find this spec proposal interesting:
open-telemetry/opentelemetry-specification#3216

[krutoo]

Сan someone tell me how to at least somehow set buckets for histogram now?

[weyert]

You need to define a View for it before the instruments are created

[weyert]

new View({
aggregation: new ExplicitBucketHistogramAggregation([1,5,10,25], true),
instrumentName: metricName,
instrumentType: InstrumentType.HISTOGRAM,
})

[krutoo]

new View({ aggregation: new ExplicitBucketHistogramAggregation([1,5,10,25], true), instrumentName: metricName, instrumentType: InstrumentType.HISTOGRAM, })

thanks but in unit test we found other way to achieve it:

opentelemetry-js/packages/sdk-metrics/test/MeterProvider.test.ts

Line 473 in 3a2ca58

it('with instrument unit should apply view to only the selected instrument unit', async () => {

Opentelemetry is really great framework but looks like some cases are too complex here

[github-actions]

This issue is stale because it has been open 60 days with no activity. Remove stale label or comment or this will be closed in 14 days.

[jdmarshall]

Hmm, I think I had this suggestion working and now it's stopped. Is the metricName required? I thought I read that for instance just a units label would be enough. (eg, bytes versus ms)