@papack/metric

TypeScript metrics for developers who want minimal runtime and application metrics.

Installation

npm install @papack/metric

Concept

The package is built around a single abstraction:

interface MetricPortInterface<K extends string = string> {
  set(key: K, value: number | ((prev: number) => number)): void;
  snapshot(): Record<K, number>;
}

Anything that collects metrics implements this interface.

At the moment two concrete implementations are provided:

• Metric - in-memory application metrics, includes no metric
• RuntimeMetric - minimal runtime and system metrics

You can swap them freely via the interface.

Metric (application metrics)

Use Metric for any number value.

Example

import { Metric } from "@papack/metric";

const metric = new Metric(["REQUESTS_TOTAL", "ERRORS_TOTAL"] as const);

metric.set("REQUESTS_TOTAL", 1);
metric.set("REQUESTS_TOTAL", (v) => v + 1);
metric.set("ERRORS_TOTAL", 0);

const snapshot = metric.snapshot();
// {
//   REQUESTS_TOTAL: 2,
//   ERRORS_TOTAL: 0
// }

Properties

• Keys are fixed and enforced at compile time
• Unknown keys throw immediately
• snapshot() always returns a full object

RuntimeMetric

RuntimeMetric collects application and OS-level metrics and allows you to extend them with custom keys.

Default Metrics

UPTIME_SECONDS;
LOOP_PRESSURE;
LOAD_1M;
LOAD_5M;
LOAD_15M;
TOTAL_MEMORY_BYTES;
FREE_MEMORY_BYTES;
HEAP_USED_BYTES;
RSS_BYTES;

Example

import { RuntimeMetric } from "@papack/metric";

const metric = new RuntimeMetric(["REQUESTS_TOTAL"]);

metric.set("REQUESTS_TOTAL", 1);

const snapshot = metric.snapshot();
// {
//   UPTIME_SECONDS: number,
//   LOOP_PRESSURE: number,
//   LOAD_1M: number,
//   LOAD_5M: number,
//   LOAD_15M: number,
//   TOTAL_MEMORY_BYTES: number,
//   FREE_MEMORY_BYTES: number,
//   HEAP_USED_BYTES: number,
//   RSS_BYTES: number,
//   REQUESTS_TOTAL: number
// }