rakomon is designed for simplest, most stupid monitoring tasks when logging, caching and other features of advanced monitoring solutions are not required.
It encapsulates some of tornado and apscheduler functionality to run simple tasks periodically, store results in memory and serve them in application/json.
You can use both monitor and endpoint modules to serve your metrics automagically, or instantiate the monitor.Monitor class on your own to use it however you see fit.
Disclaimer: though I use this code in my projects, it's rather raw. I set up this repository mainly to learn how to maintain an open-source project. It's up to you to evaluate if rakomon is production ready.
rakomon is Python 3 only. It also requires tornado and apscheduler.
Use pip:
$ pip install rakomonOr clone repository to your machine and use easy_install:
$ easy_install setup.pyrakomon is ready out of the box. All you need is get default Monitor instance, then define your metrics as functions and decorate them with the Monitor.metric decorator.
A metric function should return either a number or a string. Iterables support is planned for future.
After you define your metrics, call endpoint.run() to launch a simple tornado server to serve them. It will bind to port 80 on localhost by default.
import psutil
from rakomon import monitor, endpoint
m = monitor.default()
@m.metric
def cpu():
return psutil.cpu_percent(interval=1)
@m.metric
def ram():
return psutil.virtual_memory().percent
endpoint.run()$ curl http://127.0.0.1/rakomon
{"cpu": 2.3, "ram": 33.1}Both monitor.Monitor and endpoint.run are configurable. You can pass configuration as keyword arguments. If kwargs are not provided, rakomon uses default values.
scheduler- a class that inherits fromapscheduler.schedulers.base.BaseScheduler. Useful when you want to utilize Monitor on its own, without the included endpoint. See apscheduler docs for more info on schedulers. Defaults toTornadoScheduler.metric_interval- number of idle seconds between metric runs. Defaults to5.round_ndigits- number of digits when rounding metric values. Naturally, applies to number values only. Defaults to2.
monitor- an instance ofmonitor.Monitor. Defaults tomonitor.default().address- address to bind the tornado endpoint to. Defaults to'127.0.0.1'.port- port to bind the tornado endpoint to. Defaults to80.url_path- defaults tor'/rakomon'.