gistfile1.js

var timeseries = require('timeseries')({
   elasticsearch_hosts : [{ 
     port : 9200, 
     host : 'localhost' 
   }],
   redis : {
     port : 6379,
     host : 'localhost'
   }
});
 

/**
 * Here is the data we receive every second
 * 
 * data = {
 *    metrics : {
 *      clients : 2,
 *      events_1sec : 10,
 *      events_10sec : 100
 *    },
 *    process: {
 *      name : "auto-throw",
 *      pm_id: 11,
 *      server: "ultraio",
 *      rev: "1703a71b2d2a449f8f18ab0501e73a86f24eab2a"
 *    }
 * };
 * 
 */


/**
 * 1- The data is first aggregated every minutes (in Redis)
 *
 *    {
 *       clients     : [0, 0, 14, 15, 43, 30],
 *       events_1sec : [0, 0, 30, 50, 89, 90],
 *       events_1min : [0, 0, 130, 400, 599, 300]
 *    }
 *
 * 2- When the current minute is reached a simple average calculation is made and we append the extra_data and the minute:
 *
 *    {
 *       clients     : 25,
 *       events_1sec : 40,
 *       events_1min : 380,
 *       at          : 'Wed Sep 24 2014 17:29:00 GMT+0200 (CEST)',
 *       process     : {
 *         name    : "auto-throw",
 *         pm_id   : 11,
 *         server  : "ultraio",
 *         rev     : "1703a71b2d2a449f8f18ab0501e73a86f24eab2a"
 *       }
 *   }
 *
 * 3 - The data is then flushed to database
 * 
 */
 
timeseries.put({
  //uid          : 'app_name:id',       // Should we keep it or we can instead use the extra_data field to generate this UID? 
  index        : 'my:index:metrics', 
  extra_data   : {   // This data must be attached to the entry when save into Elasticsearch 
      process: {
         name  : "app-name",
         pm_id : 11,
         server: "server-name",
         rev   : "1703a71b2d2a449f8f18ab0501e73a86f24eab2a"
      }
  },
  // fields       : ['clients', 'events_1sec'], // fields must be automatically retrieved depending on the data we save
  data         : {
     clients     : 2,
     events_1sec : 10,
     events_1min : 65
  }
}, function(error, elementsStored){});
 

 
/**
 * The outputted data should look like this (ES automatically do this):
 *
 * {
 *  clients     : [[12312233212, 1], [12312331232, 10], [12312233212, 14]], //mean value for grouped by minute
 *  events_1sec : [[12312233212, 120], [12312331232, 120], [12312233212, 145], //mean value for grouped by minute
 *  events_1min : [[12312233212, 1900], [12312331232, 600], [12312233212, 1490], //mean value for every minute 
 * }
 *
 * fields: if it's null or empty, it should first get the most recent entry and extract the fields
 *         Then it does an aggs query on all these fields
 * filter: for each key/value we add on it, it should just add a filter parameter
 *       EG: query.body.query.filtered.filter.and.push({
 *             term: {
 *                'process.name' : opts.app_name
 *             }
 *           });
 * 
 */
timeseries.get({
  //uid          : 'app_name:id',
  index        : 'my:index:metrics',
  fields       : ['clients', 'events_1sec'],  // If this field is empty or null it should retrieve fields dynamically
  from         : 'now-14d',
  to           : 'now'
  interval     : 'hour',
  filter       : {             // These key/value append a filter
     'process.name' : 'app-name'
  }
}, function(err, data) {
  if(err){
    throw err;
  } else {
    console.log(data);
  }
});