@opentelemetry/metrics: The option "boundaries" for metrics is not propagated to the HistogramAggregator

[jhbabon]

Please answer these questions before submitting a bug report.

What version of OpenTelemetry are you using?

Version v0.12.

What version of Node are you using?

Node 14.10.1

Please provide the code you used to setup the OpenTelemetry SDK

package.json

{
  "name": "otel-bug-repro",
  "version": "1.0.0",
  "description": "OTel Metrics bug repro",
  "main": "index.js",
  "scripts": {
    "test": "echo \"Error: no test specified\" && exit 1"
  },
  "author": "",
  "license": "UNLICENSED",
  "dependencies": {
    "@opentelemetry/core": "^0.12.0",
    "@opentelemetry/exporter-prometheus": "^0.12.0",
    "@opentelemetry/metrics": "^0.12.0",
    "node-fetch": "^2.6.1"
  }
}

index.js

'use strict';

const { MeterProvider } = require('@opentelemetry/metrics');
const { ConsoleLogger, LogLevel } = require('@opentelemetry/core');
const { PrometheusExporter } = require('@opentelemetry/exporter-prometheus');
const fetch = require('node-fetch');

function main() {
  const scrapeUrl = `http://localhost:${PrometheusExporter.DEFAULT_OPTIONS.port}${PrometheusExporter.DEFAULT_OPTIONS.endpoint}`
  const logger = new ConsoleLogger(LogLevel.DEBUG)
  const exporter = new PrometheusExporter(
    {
      preventServerStart: false,
      logger,
    },
    () => {
      console.log(`prometheus scrape endpoint: ${scrapeUrl}`);
    },
  );

  const meter = new MeterProvider({
    exporter,
    interval: 1000,
    logger,
  }).getMeter('example-observer');

  const tester = meter.createValueRecorder('tester', {
    description: 'test boundaries propagation',
    logger,
    boundaries: [0.01, 5, 10], // these are not propagated to Prometheus' Histogram
  });

  tester.record(8);

  setTimeout(async () => {
    const res = await fetch(scrapeUrl);
    const metrics = await res.text();

    console.log(metrics);

    process.exit(1);
  }, 1500);
}

main();

What did you do?

I'm trying to create a Promtetheus Histogram using a ValueRecorder metric, but when setting the boundaries option this same option is not translated to the buckets list that the Histogram should be using. Somehow the value is lost in the conversion.

What did you expect to see?

I would expect to see a list of *_bucket counters with the le tag with values 0.01, 5 and 10. Something similar to this (based on the example):

# HELP tester tester
# TYPE tester histogram
tester_count 1 1603784561929
tester_sum 8 1603784561929
tester_bucket{le="0.01"} 0 0
tester_bucket{le="5"} 0 0
tester_bucket{le="10"} 1 1603784561929
tester_bucket{le="+Inf"} 1 1603784561929

What did you see instead?

If a run the example you can see the buckets are not set up:

$ node index.js
node index.js
Prometheus exporter started on port 9464 at endpoint /metrics
prometheus scrape endpoint: http://localhost:9464/metrics
Prometheus exporter export
# HELP tester tester
# TYPE tester histogram
tester_count 1 1603784561929
tester_sum 8 1603784561929
tester_bucket{le="Infinity"} 1 1603784561929
tester_bucket{le="+Inf"} 1 1603784561929

[AndrewGrachov]

will take it

[vmarchaud]

@AndrewGrachov Feel free to reach out if you have any questions

[AndrewGrachov]

OK, I see this is my "favorite" thing with otel :)

#1477

Guess we should remove histogram option from createValueRecorder metric options

[AndrewGrachov]

@jhbabon pls see #1465

[AndrewGrachov]

The PR above should fix the issue, but I'm not sure if its logically correct to specify histogram-aggregator specific parameters within Base Metric class (which could be possibly aggregated many ways)

[jhbabon]

The PR above should fix the issue, but I'm not sure if its logically correct to specify histogram-aggregator specific parameters within Base Metric class (which could be possibly aggregated many ways)

Many thanks!