"Graphite is actually a bit of a niche application. Specifically, it is designed to handle numeric time-series data. For example, Graphite would be good at graphing stock prices because they are numbers that change over time." [http://graphite.wikidot.com/faq#toc0]

YAG is yet another graphite. I started this project mostly because I wanted to learn something... Current version is a minimalistic implementation of graphite's engine (yes, just engine - for rendering charts I recommand 3rd party tool - giraffe). Right now supports only basic API functions like: sumSeries, divSeries and diffSeries.

YAG contains 3 independent components which follow Unix philosophy - "do one thing and do it well":

• listener - stores datapoints received from "StatsD server"
• webserver - serves datapoints for dashboard clients (implements graphite json rest api)
• ttl - daemon which removes datapoints older than set number of seconds

For metrics management I recommand phpRedisAdmin. Here is my fork a little bit modified:

https://github.com/kuba--/phpRedisAdmin/tree/yag

For metrics dashboard, I recommand giraffe. Here is my fork:

https://github.com/kuba--/giraffe

• YAG was implemented in "Go", so install golang first [http://golang.org/doc/install].

• YAG uses redis database (check redis.conf for configuration details) to store datapoints, so you need to install redis database [http://redis.io/download] on your DB server. Redis version >= 2.6.0 required in order to load lua scripts

• Set up $GOPATH (e.g.: $GOPATH=$HOME/workspace).

• Now you can compile YAG's code. Redis' driver (redix) and glog library will be installed automatically.

    cd $GOPATH
  
    go get github.com/kuba--/yag/listener
    go get github.com/kuba--/yag/webserver
    go get github.com/kuba--/yag/ttl
  

• If you already downloaded/cloned a code, you would be able to use make command-line tool.

    # just compile
    make
  
    # check deploy target if you want to compile and deploy
    make deploy
    
    # if you installed golang with cross-compile flag, you would be able to compile yag for any platform (e.g. linux)
    make -e GOOS=linux GOARCH=amd64 
  

• Executable files are in $GOPATH/bin directory.

• You can run listener, webserver and ttl with following flags:

   -f="config.json": path to the config file
  
   -alsologtostderr=false: log to standard error as well as files
   -log_dir="": If non-empty, write log files in this directory
   -logtostderr=false: log to standard error instead of files
   -stderrthreshold=0: logs at or above this threshold go to stderr
  

• Example:

    $ ./listener -f=./config.json -log_dir=./logs -stderrthreshold=INFO
  

• You can also profiling listener and webserver adding following flags:

    -cpuprofile="": write cpu profile to file
    -memprofile="": write memory profile to this file
  

More about profiling go programs you can find here: http://blog.golang.org/profiling-go-programs

• Configuration file (e.g. config.json):

Note 1: remove comments before using this config

{
	"DB":{                          // Database section
		"Addr":"localhost:6379",     // address and port of Redis DB
		"Timeout":30,                // timeout per connection (in seconds)
		"MaxClients":30              // maximum number of clients in DB connection pool
	},
	"Metrics":{                     // Metrics section
		"GetScript":"get.lua",       // relative path to get script
		"AddScript":"add.lua",       // relative path to add script					
		"TtlScript":"ttl.lua",       // relative path to ttl script
		"TTL":86600,                 // time to live per metric (in seconds)
		"ConsolidationStep":60,      // consolidate datapoints per step (in seconds).
		"ConsolidationFunc":"avg"    // function used to consolidate datapoints (valid function names: sum, avg, min, max)
	},
	"Listener":{                    // Listener server section 
		"Addr":":2003"               // local address and port
	},
	"Webserver":{                   // Webserver section
		"Addr":":8080",              // local address and port
		"Timeout":30                 // timeout per connection (in seconds)
	},
	"TTL":{                         // TTL daemon section
		"Tick":12                    // timers tick (in seconds)
	}
}

Note 2 (Webserver): if you remove properties: "ConsolidationStep", "ConsolidationFunc" from config file, webserver will not consolidate datapoints.

Note 3 (Webserver): if you add "maxDataPoints" parameter > 0 then "ConsolidationStep" can be changed by webserver to return around "maxDataPoints". For instance, for following data points received from StatsD daemon:

[0,1384613389],[0,1384613399],[0,1384613409],[0.5,1384613419],[0.75,1384614209]

with ConsolidationFunc: "avg" and maxDataPoints=7, webserver will output following data points:

[0.12, 1384613389]
[null, 1384613509]
[null, 1384613629]
[null, 1384613749]
[null, 1384613869]
[null, 1384613989]
[0.75, 1384614109]

Copyright 2013 Kuba Podgorski

Unless otherwise noted, the source files are distributed under the MIT License found in the LICENSE file.