influx-collector

InfluxDB stats collector for node.js

For InfluxDB 0.9.x use version 1.x.x of this module.

For InfluxDB <= 0.8.x use version 0.x.x of this module.

Note that the API changed between 0.x and 1.x to move the series name into the constructor.

quickstart

// set some env var (i.e INFLUX_STATS_URI)
// http://user:password@influxdb.host.com/stats-db;

var Stats = require('influx-collector');

// create a stats collector
var mem_stats = Stats('process-memory-usage', process.env.INFLUX_STATS_URI);

// example stat collection of process memory usage
setInterval(function() {
    var mem = process.memoryUsage();

    // collect this stat into the 'process-memory-usage' series
    // first argument is the data points
    // second (optional) argument are the tags
    mem_stats.collect({
        rss: mem.rss,
        heap_total: mem.heapTotal,
        heap_used: mem.heapUsed
    }, {
        pid: process.pid,
        app: 'my-app'
    });
}, 10 * 1000);

// make sure to handle errors to avoid uncaughtException
// would be annoying if stats crashed your app :)
mem_stats.on('error', function(err) {
    console.error(err); // or whatever you want
});

Options

influx-collector is configured via the URI string. The following options can be set by specifying them as query parameters in the URI

Param	Default	Description
autoFlush	yes	Specify `no` to avoid stats flushing automatically. See `flushing` below.
flushInterval	5000	Stats are flushed every flushInterval milliseconds.
instantFlush	no	Flush stats to server on collection.
time_precision		Time precision when writing data. Only relevant if you are specifying the `time` field.

Flushing and Batches

By default, stats are flushed in batches every flushInterval milliseconds. This allows for many stats to be collected and flushed at once. You can override these behaviors via the autoFlush and interval options as query parameters. You can also turn on instantFlush which will send the stats to InfluxDB as soon as they are collected (every collect call will be an http request). I do not recommend this approach for high frequency stats.

Example of a 10 second flush interval

http://user:password@influxdb.host.com/stats-db?flushInterval=10000

Stat#flush()

To manually flush the stats at any time (or if autoFlush is off), use the .flush method on your stats instance.

Protips

Make sure to attach and error handler so stats don't crash your entire app.
Useful to log pid and app names/tags for generic stats so you can group by later.
Grafana is amazing.

malisper/node-influx-collector