devmetrics-core

devmetrics-nodejs-core

NodeJS lib to instrument application for logs, metrics and usage events in a consistent way.

Library supports nice stdout formatting and log levels.

Events are also sent to a logstash over UDP. Use your logstash instance or the devmetrics.io shared ELK (elasticsearch, logstash, kibana) instance.

Browse data in the Kibana, Grafana or native interface at http://www.devmetrics.io/logslib

Installation

  $ npm install devmetrics-core

Library init

Default way, nice stdout logs and free devmetrics.io shared cluster:

var logger = require('devmetrics-core')();
logger.info('hello world');

Advanced init (for example, to send data to your private instance):

var devmetrics = require('devmetrics-core')({
  'host': 'your-hostname',
  'port': 5545,
  'app_id': 'my-private-id',
  'no_console': false,
  'software_version': '1.2'
});

All settings are optional:

• host - Set to your private instance of logstash, if required, defaults to service.devmetrics.io
• port - Logstash port, defaults to 5546 for devmetrics.io cluster
• app_id - The unique app_id to filter your data and access dashboards, defaults to os.hostname()
• no_console - Set to true for the silent mode (no stdout), defaults to false
• software_version - Your app's version. Useful for continuous integration, defaults to ``

Simple Logs API

Simple log

devmetrics.info('hey info, not important');
devmetrics.warn('hey warn, probably important');
devmetrics.error('hey error, really important');

Basic logging, for anything. Levels: trace, debug, info, warn, error

Exception

devmetrics.exception(e);

Adds special sections to track application problems

Event API

User event

/**
 * @param event_name Metric name, all events with same metric name are aggregated
 * @param user_id optional, User unique identifier, used only for log
 * @param event_tags optional, Additional info, metric dimensions, string or array
 */
devmetrics.userEvent('purchase_button_click', 123, ['android', 'US']);
devmetrics.userEvent('login', 89321);

We are generally user oriented and track user actions. So let's separate business metrics from developer data.

App event

/**
 * @param event_name Metric name, all events with same metric name are aggregated
 * @param event_tags optional, Additional info, metric dimensions, string or array
 */
devmetrics.appEvent('app_started', 'server2');
devmetrics.appEvent('web-request', ['index-page', 'US_region']);

Almost the same API as for UserEvent, but userEvent is for business metrics and appEvent is for application and system metrics. Let's differentiate this data starting from the collect layer.

Code performance measurement

To measure a specific function execution time, you can wrap it with the devmetrics function wrapper. This enables you to see how often your function is called and how many milliseconds it takes to execute.

var myProbablySlowFunc = function () {...};
myProbablySlowFunc = devmetrics.measureTimeWrap(myProbablySlowFunc, 'SomeNameForDashboard');

Out of the box instrumentation, Application Performance Monitoring

For MEAN stack we've prepared easy-to-go methods for APM instrumentation:

devmetrics.enableHttpLogger(app); // Express.js HTTP requests
devmetrics.enableMongooseLogger(mongoose); // MongoDB calls instrumentation

The idea: logs + metrics in one request

There are a lot of questions we need to answer about our app every day: business metrics for a product, application metrics for developers, system metrics for devops.

Metrics or logs? Right tool, right job.

• Metrics are to provide aggregated information generally visualized as graphs.
• Logs are for a detailed drilldown. View specific user actions or debug code behaviour.

Our lib api design relies on several principles:

1. Metrics and logs go together, we provide a single api to write both logs and metrics. For each event the API produces metrics with aggregated values and log entry with event details.

2. For high volume applications the log entries can be sampled. The combined log and metrics approach allows to preserve accurate aggregated information in metrics while significantly reducing diagnostic log volume.

3. Tracking user acivity is a key requirement for modern apps. We provide userEvent as first class citizen in the libary to simplify user activity tracking. In addition to userEvent we also provide appEvent to track structured information about application levels events such as start, stop, etc.

Checkout chat with out development team for comments and questions: https://gitter.im/devmetrics/dev#

Stay in touch

We are moving fast and appreciate feedback and feature requests. See you @ https://gitter.im/devmetrics/dev#