perf: improve performance of rendering metrics to Prometheus string (#549)

Author: ngavalas

CHANGELOG.md

@@ -13,6 +13,8 @@ project adheres to [Semantic Versioning](http://semver.org/).
 
 ### Changed
 
+- Refactor histogram internals and provide a fast path for rendering metrics to
+  Prometheus strings when there are many labels shared across different values.
 - Disable custom content encoding for pushgateway delete requests in order to
   avoid failures from the server when using `Content-Encoding: gzip` header.
 - Refactor `escapeString` helper in `lib/registry.js` to improve performance and

lib/histogram.js

@@ -97,6 +97,12 @@ class Histogram extends Metric {
 	}
 
 	async get() {
+		const data = await this.getForPromString();
+		data.values = data.values.map(splayLabels);
+		return data;
+	}
+
+	async getForPromString() {
 		if (this.collect) {
 			const v = this.collect();
 			if (v instanceof Promise) await v;
@@ -176,9 +182,10 @@ function startTimer(startLabels) {
 	};
 }
 
-function setValuePair(labels, value, metricName, exemplar) {
+function setValuePair(labels, value, metricName, exemplar, sharedLabels = {}) {
 	return {
 		labels,
+		sharedLabels,
 		value,
 		metricName,
 		exemplar,
@@ -255,20 +262,17 @@ function convertLabelsAndValues(labels, value) {
 function extractBucketValuesForExport(histogram) {
 	return bucketData => {
 		const buckets = [];
-		const bucketLabelNames = Object.keys(bucketData.labels);
 		let acc = 0;
 		for (const upperBound of histogram.upperBounds) {
 			acc += bucketData.bucketValues[upperBound];
 			const lbls = { le: upperBound };
-			for (const labelName of bucketLabelNames) {
-				lbls[labelName] = bucketData.labels[labelName];
-			}
 			buckets.push(
 				setValuePair(
 					lbls,
 					acc,
 					`${histogram.name}_bucket`,
 					bucketData.bucketExemplars[upperBound],
+					bucketData.labels,
 				),
 			);
 		}
@@ -281,21 +285,40 @@ function addSumAndCountForExport(histogram) {
 		acc.push(...d.buckets);
 
 		const infLabel = { le: '+Inf' };
-		for (const label of Object.keys(d.data.labels)) {
-			infLabel[label] = d.data.labels[label];
-		}
 		acc.push(
 			setValuePair(
 				infLabel,
 				d.data.count,
 				`${histogram.name}_bucket`,
 				d.data.bucketExemplars['-1'],
+				d.data.labels,
+			),
+			setValuePair(
+				{},
+				d.data.sum,
+				`${histogram.name}_sum`,
+				undefined,
+				d.data.labels,
+			),
+			setValuePair(
+				{},
+				d.data.count,
+				`${histogram.name}_count`,
+				undefined,
+				d.data.labels,
 			),
-			setValuePair(d.data.labels, d.data.sum, `${histogram.name}_sum`),
-			setValuePair(d.data.labels, d.data.count, `${histogram.name}_count`),
 		);
 		return acc;
 	};
 }
 
+function splayLabels(bucket) {
+	const { sharedLabels, labels, ...newBucket } = bucket;
+	for (const label of Object.keys(sharedLabels)) {
+		labels[label] = sharedLabels[label];
+	}
+	newBucket.labels = labels;
+	return newBucket;
+}
+
 module.exports = Histogram;

lib/registry.js

@@ -29,7 +29,10 @@ class Registry {
 	}
 
 	async getMetricsAsString(metrics) {
-		const metric = await metrics.get();
+		const metric =
+			typeof metrics.getForPromString === 'function'
+				? await metrics.getForPromString()
+				: await metrics.get();
 
 		const name = escapeString(metric.name);
 		const help = `# HELP ${name} ${escapeString(metric.help)}`;
@@ -41,6 +44,7 @@ class Registry {
 
 		for (const val of metric.values || []) {
 			let { metricName = name, labels = {} } = val;
+			const { sharedLabels = {} } = val;
 			if (
 				this.contentType === Registry.OPENMETRICS_CONTENT_TYPE &&
 				metric.type === 'counter'
@@ -52,11 +56,19 @@ class Registry {
 				labels = { ...labels, ...defaultLabels, ...labels };
 			}
 
-			const formattedLabels = formatLabels(labels);
-			const labelsString = formattedLabels.length
-				? `{${formattedLabels.join(',')}}`
-				: '';
+			// We have to flatten these separately to avoid duplicate labels appearing
+			// between the base labels and the shared labels
+			const formattedLabels = [];
+			for (const [n, v] of Object.entries(labels)) {
+				if (Object.prototype.hasOwnProperty.call(sharedLabels, n)) {
+					continue;
+				}
+				formattedLabels.push(`${n}="${escapeLabelValue(v)}"`);
+			}
 
+			const flattenedShared = flattenSharedLabels(sharedLabels);
+			const labelParts = [...formattedLabels, flattenedShared].filter(Boolean);
+			const labelsString = labelParts.length ? `{${labelParts.join(',')}}` : '';
 			values.push(
 				`${metricName}${labelsString} ${getValueAsString(val.value)}`,
 			);
@@ -207,6 +219,18 @@ function formatLabels(labels) {
 	);
 }
 
+const sharedLabelCache = new WeakMap();
+function flattenSharedLabels(labels) {
+	const cached = sharedLabelCache.get(labels);
+	if (cached) {
+		return cached;
+	}
+
+	const formattedLabels = formatLabels(labels);
+	const flattened = formattedLabels.join(',');
+	sharedLabelCache.set(labels, flattened);
+	return flattened;
+}
 function escapeLabelValue(str) {
 	if (typeof str !== 'string') {
 		return str;

test/registerTest.js

@@ -505,15 +505,15 @@ describe('Register', () => {
 						const metrics = await r.metrics();
 						const lines = metrics.split('\n');
 						expect(lines).toContain(
-							'my_histogram_bucket{le="1",type="myType",env="development"} 1',
+							'my_histogram_bucket{le="1",env="development",type="myType"} 1',
 						);
 
 						myHist.observe(1);
 
 						const metrics2 = await r.metrics();
 						const lines2 = metrics2.split('\n');
 						expect(lines2).toContain(
-							'my_histogram_bucket{le="1",type="myType",env="development"} 2',
+							'my_histogram_bucket{le="1",env="development",type="myType"} 2',
 						);
 					});
 				});