Tracking the weather with Python and Prometheus

Create a custom Prometheus integration to keep track of the biggest cloud provider of all: Mother Earth.
164 readers like this
164 readers like this
14 facts about OpenStack Newton

Pixabay. Modified by CC BY-SA 4.0

Open source monitoring system Prometheus has integrations to track many types of time-series data, but if you want an integration that doesn't yet exist, it's easy to build one. An often-used example is a custom integration with a cloud provider that uses the provider's APIs to grab specific metrics. In this example, though, we will integrate with the biggest cloud provider of all: Earth.

Luckily, the US government already measures the weather and provides an easy API for integrations. Getting the weather forecast for the next hour at Red Hat headquarters is simple.

import requests
HOURLY_RED_HAT = ",57/forecast/hourly"
def get_temperature():
    result = requests.get(HOURLY_RED_HAT)
    return result.json()["properties"]["periods"][0]["temperature"]

Now that our integration with Earth is done, it's time to make sure Prometheus can understand what we are saying. We can use the Prometheus Python library to create a registry with one gauge: the temperature at Red Hat HQ.

from prometheus_client import CollectorRegistry, Gauge
def prometheus_temperature(num):
    registry = CollectorRegistry()
    g = Gauge("red_hat_temp", "Temperature at Red Hat HQ", registry=registry)
    return registry

Finally, we need to connect this to Prometheus in some way. That depends a little on the network topology for Prometheus: whether it is easier for Prometheus to talk to our service, or whether the reverse is easier.

The first case is the one usually recommended, if possible, so we need to build a web server exposing the registry and then configure Prometheus to scrape it.

We can build a simple web server with Pyramid.

from pyramid.config import Configurator
from pyramid.response import Response
from prometheus_client import generate_latest, CONTENT_TYPE_LATEST
def metrics_web(request):
    registry = prometheus_temperature(get_temperature())
    return Response(generate_latest(registry),
config = Configurator()
config.add_route('metrics', '/metrics')
config.add_view(metrics_web, route_name='metrics')
app = config.make_wsgi_app()

This can be run with any Web Server Gateway Interface (WSGI) server. For example, we can use python -m twisted web --wsgi to run it, assuming we put the code in

Alternatively, if it is easier for our code to connect to Prometheus, we can push it to Prometheus's Push gateway periodically.

import time
from prometheus_client import push_to_gateway
def push_temperature(url):
    while True:
        registry = prometheus_temperature(get_temperature())
        push_to_gateway(url, "temperature collector", registry)

The URL is the one for the Push gateway; it often ends in :9091.

Good luck building your own custom Prometheus integration so you can track all the things!

Moshe sitting down, head slightly to the side. His t-shirt has Guardians of the Galaxy silhoutes against a background of sound visualization bars.
Moshe has been involved in the Linux community since 1998, helping in Linux "installation parties". He has been programming Python since 1999, and has contributed to the core Python interpreter. Moshe has been a DevOps/SRE since before those terms existed, caring deeply about software reliability, build reproducibility and other such things.

1 Comment

Creative Commons LicenseThis work is licensed under a Creative Commons Attribution-Share Alike 4.0 International License.