DAZN Datadog Metrics

Helper module for recording metrics with Datadog. It supports both synchronous (via HTTP) and asynchronous (via Datadog's Lambda integration) modes. Defaults to asynchronous mode.

Main features:

• Supports both sync and async methods of collecting metrics. Mode can be configured via the DATADOG_METRICS_MODE environment variable, use sync for sync mode, async for async mode

• Supports gauge, histogram and increment

• Supports tracking of a function's execution time (see below for more details)

• Support namespacing. You can specify prefix for all metrics with the DATADOG_PREFIX environment variable

• Support default tags. You can specify default tags for all metrics with the DATADOG_TAGS environment variable, which should be a comma-separated value, e.g. environment:dev,awsRegion:us-east-1,load_test

Configuration options (environment variables):

• DATADOG_METRICS_MODE: async | sync

• DATADOG_PREFIX: string (optional), used to provide namespacing, e.g. giftcodes.

• DATADOG_TAGS: string (optional), default tags in comma-separate format, e.g. environment:dev,awsRegion:us-east-1,load_test

Getting Started

Install from NPM: npm install @dazn/datadog-metrics

API

Basic usage looks like this:

// if process.env.DATADOG_METRICS_MODE === 'sync' then this will be recording
// metrics with the 'datadog-metrics' package, and would require flushing
// you can use @perform/lambda-powertools-middleware-flush-datadog-metrics to
// take care of that
// if process.env.DATADOG_METRICS_MODE === 'async' then metrics would be
// sent to stdout, using the DogStatsD format
const Datadog = require('@dazn/datadog-metrics')

Datadog.gauge('key', 42)                        // guage without tags
Datadog.gauge('key', 42, [ 'tag1', 'tag2:42' ]) // gauge with tags
Datadog.gauge('key', 42, [], Date.now())        // guage with timestamp override

Datadog.increment('key')                            // increment by 1, no tags
Datadog.increment('key', 42)                        // increment by 42, no tags
Datadog.increment('key', 42, [ 'tag1', 'tag2:42' ]) // increment by 42, with tags
Datadog.increment('key', 42, [], Date.now())        // increment with timestamp override

Datadog.histogram('key', 42)                        // histogram without tags
Datadog.histogram('key', 42, [ 'tag1', 'tag2:42' ]) // histogram with tags
Datadog.histogram('key', 42, [], Date.now())        // histogram with timestamp override

const f = () => {
  // ...
}

Datadog.trackExecTime(f, 'key')                        // track execution time without tags
Datadog.trackExecTime(f, 'key', [ 'tag1', 'tag2:42' ]) // track execution time with tags
// by default, timestamp of the metric is the time when the function finishes, but you can
// override the timestamp yourself
Datadog.trackExecTime(f, 'key', [], Date.now())

const asyncF = async () => {
  // ...
}

// track async function execution time without tags
Datadog.trackExecTime(asyncF, 'key')

// track async function execution time with tags
Datadog.trackExecTime(asyncF, 'key', [ 'tag1', 'tag2:42' ])

// track async function execution time with timestamp override
Datadog.trackExecTime(asyncF, 'key', [], Date.now())

Disabling console output

Sometimes you won't want the console output to be displayed (e.g. when running your unit tests) when running in async mode. In order to do this, you can set the environment variable DISABLE_METRICS=true; this will disable console logging output. Note: Only set this for scenarios you don't want metrics to be recorded, as setting this will stop any metrics from being recorded.

For example, your scripts section in package.json may look like this:

  "scripts": {
    "unit:test": "DISABLE_METRICS=true jest",
    ...
  }