Skip to main content

Exposes applipy metrics in prometheus format

Project description

pipeline status coverage report PyPI Status PyPI Version PyPI Python PyPI License PyPI Format

Applipy Prometheus Metrics

pip install applipy_prometheus

Exposes applipy metrics in prometheus format as an HTTP endpoint with path /metrics.

Usage

Add the applipy_prometheus.PrometheusModule to your application. Optionally, define through which http server to expose the /metrics endpoint, if no name is given it defaults to the anonymous server:

# dev.yaml

app:
  name: demo
  modules:
    - applipy_prometheus.PrometheusModule

http.servers:
- name: internal
  host: 0.0.0.0
  port: 8080

prometheus.server_name: internal

To run this test just install applipy_prometheus and pyyaml and run the applipy application:

pip install applipy_prometheus pyyaml
python -m applipy

You can now query http://0.0.0.0:8080/metrics and you should see some metrics for that endpoint (you'll have to query it twice to see metrics).

This module uses applipy_metrics's registry to load the metrics and generate the Prometheus document.

Metrics Endpoint Wrapper

This library also comes with MetricsWrapper. It is an applipy_http.EndpointWrapper that can be bound to your APIs and will automatically measure the request time and store it as a summary with name applipy_web_request_duration_seconds.

By default, the library will add the MetricsWrapper to the API that registers the endpoint /metrics, named prometheus, and the anonymous API, the one that is bound without named parameters. This functionalities can be disabled by setting the configuration values prometheus.observe_prometheus_api and prometheus.observe_anonymous_api to false.

A named API is one that is registered with named parameters like so: bind(with_names(Api, 'api_name')).
The anonymous API has no named parameters and is usually registered like so: bind(Api).

You can also tell the module to add the MetricsWrapper to you named APIs by setting the configuration value prometheus.api_names to a list containing the names of you named APIs.

In the case your API is registered with multiple parameter names, the one that applies for the wrapper is the name for the wrappers parameter.

The wrapper has priority 100.

The metrics are tagged by default with:

  • method: HTTP request method (i.e. GET, POST, etc.)
  • path: path of the endpoint handling the request
  • server: name of the server handling the request (anonymous server is empty string)
  • status: status code of the response

On top of that, a dictionary is added to the Context with the key metrics.tags where you can add custom tags to the metric.

Example

Full prometheus module config

All keys and their default values:

prometheus:
  server_name: null
  observe_prometheus_api: true
  observe_anonymous_api: true
  api_names: []

Endpoint with custom metric tag

# myendpoint.py

from aiohttp import web
from applipy_http import Endpoint


class MyEndpoint(Endpoint):

    async def get(self, req, ctx):
        ctx['metrics.tags']['custom_tag'] = 'value'
        return web.Response(body='Ok')

    def path(self):
        return '/'

Usage with anonymous API

# mymodule.py

from applipy import Module
from applipy_http import Api, HttpModule, Endpoint, PathFormatter
from applipy_inject import with_names
from applipy_prometheus import MetricsWrapper
from myendpoint import MyEndpoint


class MyModule(Module):
    def configure(self, bind, register):
        bind(Endpoint, MyEndpoint)
        bind(PathFormatter)
        bind(Api)

    @classmethod
    def depends_on(cls):
        return HttpModule,
# dev.yaml

app:
  name: test
  modules: [mymodule.MyModule]

http.servers:
- host: 0.0.0.0
  port: 8080

Usage with named API

# mymodule.py

from applipy import Module
from applipy_http import Api, HttpModule, Endpoint, PathFormatter
from applipy_inject import with_names
from applipy_prometheus import MetricsWrapper
from myendpoint import MyEndpoint


class MyModule(Module):
    def configure(self, bind, register):
        bind(Endpoint, MyEndpoint, name='myApi')
        bind(PathFormatter, name='myApi')
        bind(with_names(Api, 'myApi'))

    @classmethod
    def depends_on(cls):
        return HttpModule,
# dev.yaml

app:
  name: test
  modules: [mymodule.MyModule]

http.servers:
- host: 0.0.0.0
  port: 8080

prometheus:
  api_names: [myApi]

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

applipy_prometheus-1.1.3.tar.gz (6.7 kB view hashes)

Uploaded Source

Built Distribution

applipy_prometheus-1.1.3-py3-none-any.whl (6.6 kB view hashes)

Uploaded Python 3

Supported by

AWS AWS Cloud computing and Security Sponsor Datadog Datadog Monitoring Fastly Fastly CDN Google Google Download Analytics Microsoft Microsoft PSF Sponsor Pingdom Pingdom Monitoring Sentry Sentry Error logging StatusPage StatusPage Status page