pino-debug

pino-debug

High performance debug logging.

Seamlessly integrates the debug module with the high performance pino logger so you can turn on debug logs in production scenarios with minimum overhead.

• Up to 10x faster than using debug (20x in extreme mode!)
• JSON output with more detail (pino/bunyan/bole format)
• Safe with circular references (debug isn't)
• No need to replace any debug logging calls
• Associate namespaces with log levels
• Compatible with the entire pino ecosystem

Installation

$ npm install --save pino-debug

Usage

Preload

If all you want is fast JSON logging to STDOUT

$ DEBUG=* node -r pino-debug app.js

Namespaces are enabled the usual way, via the DEBUG environment variable.

The namespace is also included in the log output, in the ns key.

Here's a sample log when the above is applied to a generic express app:

{"pid":8784,"hostname":"Davids-MacBook-Pro.local","level":20,"time":1480277659273,"msg":"skip empty body","ns":"body-parser:json","v":1}

Programmatic

For fine grained control over output stream, and mappings between debug namespaces and pino logger levels, supply a pino instance and an optional options object with a map property containing mappings.

NOTE: pino-debug must be required at the entry point of your node process, before any other modules have been loaded

Again this example assumes a generic express app:

const pinoDebug = require('pino-debug')
const logger = require('pino')({level: process.env.LEVEL || 'info'}, process.stderr);
pinoDebug(logger, {
  auto: true, // default
  map: {
    'example:server': 'info',
    'express:router': 'debug',
    '*': 'trace' // everything else - trace
  }
})

The auto option turns on any namespaces listed in the map object (so we don't have to use the DEBUG environment variable to turn them on).

API

NOTE: pino-debug can only be called once.

pinoDebug(pinoInstance) => undefined

Call pino-debug with a pino logger instance only and any debug namespaces enabled via DEBUG or debug.enable will be logged with the level 20 ('debug').

Remember, if you want to see the messages you need to set the pino logger instance logging level to 'debug'.

pinoDebug() => undefined

Call pino-debug without arguments and a default pino instance will be created with the logging level set to 20 ('debug' level).

Any debug namespaces enabled via DEBUG or debug.enable will be logged with the level 20 ('debug').

pinoDebug(pinoInstance, opts) => undefined

This is the recommended usage. Call pino-debug with a pino logger instance, and an opts object containining map property.

opts.map {'debug-namespace: 'pino-loglevel-label'}

The keys of the map property correspond to the same namespaces that can be set on the DEBUG environment variable:

pinoDebug(pinoInstance, {
  map: {
    'my-app': 'info',
    'some-dep:*': 'debug',
    '*': 'trace'
  }
})

opts.auto [true] | false

If true (default) any debug namespaces found in the keys of opts.map will be enabled.

Additionally, any debug namespaces enabled via DEBUG or debug.enable will be logged with the level 20 ('debug').

If false, any namespaces that appear in opts.map and are enabled via DEBUG or debug.enable will be logged to with the corresponding log level, (as specified in the opts.map). Any not specified in opts.map, but which are enabled via DEBUG or debug.enable will be logged with the level 20 ('debug').

opts.skip Array

Equivalent of prefixing a namespace with dash (-) when specifying DEBUG namespaces. Any namespaces specified will not be logged.

Benchmarks

$ npm run bench

==========
basic averages
Pino average: 249
Debug average: 395
PinoDebug average: 244
PinoExtremeDebug average: 119
==========
==========
object averages
PinoObj average: 262
DebugObj average: 2448
PinoDebugObj average: 256
PinoExtremeDebugDeepObj average: 126
==========
==========
deepobject averages
PinoDeepObj average: 4809
DebugDeepObj average: 30083
PinoDebugDeepObj average: 4793
PinoExtremeDebugDeepObj average: 4810
==========

Example Folder

The example folder has a generic express app, with some additions.

The package.json file has the following scripts:

  "start": "node ./bin/www",
  "start-preload": "DEBUG=* node -r ../ ./bin/www",
  "start-programmatic": "./bin/www-programmatic",
  "start-programmatic-debug": "LEVEL=debug ./bin/www-programmatic",
  "start-programmatic-trace": "LEVEL=trace ./bin/www-programmatic"

The start-preload script demonstrates preload usage. It set's the DEBUG environment variable to log everything, and then uses the -r flag to load pino-debug (relatively referenced).

The three scripts beginning start-programmatic all use a different entry point where pino-debug has been required and instantiatd with a pino instance and the mappings (as shown in usage examples).

License

MIT

Acknowledgements

Sponsored by nearForm