cloudwatch: Unrecognized statistic warning when using Stats helper.

[everett1992]

Describe the bug

CDK can generate Unrecognized statistic warnings even when using the Stats helper.

Expected Behavior

The warning says to use the stats helper, and the helper is (somtimes, more on that later) generating valid statistics so there should not be any warnings.

Current Behavior

  new Metric({
    namespace: "example",
    metricName: "example",
    statistic: Stats.percentileRank(123),
  }).warningsV2

Unrecognized statistic "PR(:0)" for metric with namespace "example" and metric name "example". Preferably use the `aws_cloudwatch.Stats` helper class to specify a statistic. You can ignore this warning if your statistic is valid but not yet supported by the `aws_cloudwatch.Stats` helper class.

• PR(:0) is a valid statistic
• It was generated with the Stats helper class.

Reproduction Steps

import { Metric, Stats } from 'aws-cdk-lib/aws-cloudwatch';

// Stats can create valid percentile ranks but Metric will incorrectly warn on them.
it.each([
  Stats.percentileRank(0),
  Stats.percentileRank(0, 1),
  Stats.percentileRank(0, undefined),
])('Stats can create valid statistics %s that causes warnings', (statistic) => {
  const metric = new Metric({
    namespace: "example",
    metricName: "example",
    statistic,
  })

  expect(metric.warningsV2).toEqual({
    "CloudWatch:Alarm:UnrecognizedStatistic": expect.stringContaining(`Unrecognized statistic \"${statistic}\"`),
  })
});

Additionally while I was writing this bug report I realized that Stats can generate actually invalid statitsics without a type error if you pass Infinity, or NaN.

// Stats can create some statistics that are not valid, mainly because TypeScript's number
// includes special floating point values, NaN, Infinity, ect.
// Stats creates these and Metric correctly (but accidenttaly) warns on them.
it.each([
  Stats.percentileRank(0, Infinity),
  Stats.percentileRank(0, NaN),
  Stats.percentileRank(NaN),
  Stats.percentileRank(NaN),
])('Stats can create _invalid_ statistics %s that causes warnings', (statistic) => {
  const metric = new Metric({
    namespace: "example",
    metricName: "example",
    statistic,
  })

  expect(metric.warningsV2).toEqual({
    "CloudWatch:Alarm:UnrecognizedStatistic": expect.stringContaining(`Unrecognized statistic \"${statistic}\"`),
  })
});

Possible Solution

No response

Additional Information/Context

No response

CDK CLI Version

2.132.1 (build 9df7dd3)

Framework Version

2.132.1

Node.js Version

v20.10.0

OS

Ubuntu

Language

TypeScript

Language Version

No response

Other information

No response

[everett1992]

I also found that there's no way to generate open upper bounds, like PR(1:), or trimmed stats that used a fixed value, like TC(10:) vs TC(10%:), that's probably a seprate feature request.

[pahud]

yes we probably should fix here

aws-cdk/packages/aws-cdk-lib/aws-cloudwatch/lib/metric.ts

Lines 307 to 321 in 00e8a7b

	const parsedStat = parseStatistic(props.statistic || Stats.AVERAGE);
	if (parsedStat.type === 'generic') {
	// Unrecognized statistic, do not throw, just warn
	// There may be a new statistic that this lib does not support yet
	const label = props.label ? `, label "${props.label}"`: '';
	const warning = `Unrecognized statistic "${props.statistic}" for metric with namespace "${props.namespace}"${label} and metric name "${props.metricName}".` +
	' Preferably use the `aws_cloudwatch.Stats` helper class to specify a statistic.' +
	' You can ignore this warning if your statistic is valid but not yet supported by the `aws_cloudwatch.Stats` helper class.';
	this.warningsV2 = {
	'CloudWatch:Alarm:UnrecognizedStatistic': warning,
	};
	this.warnings = [warning];
	}
	this.statistic = normalizeStatistic(parsedStat);

[xazhao]

Had a quick look the percentileRank statistic is missing from here:
https://github.com/aws/aws-cdk/blob/main/packages/aws-cdk-lib/aws-cloudwatch/lib/private/statistic.ts#L187

Adding the percentileRank should fix the issue. I will draft a PR later

[github-actions]

⚠️COMMENT VISIBILITY WARNING⚠️

Comments on closed issues are hard for our team to see.
If you need more assistance, please either tag a team member or open a new issue that references this one.
If you wish to keep having a conversation with other community members under this issue feel free to do so.