Prometheus python client

Python 3 client library for Prometheus that can serve data to prometheus (in text and protobuf formats) and also push data to a pushgateway.

Table of Contents

• Prometheus python client
  • Status
  • Install
  • Why Python 3 and not python 2?
  • Usage
    • Serve data
    • Push data (to pushgateway)
  • Metrics/Collectors
    • Counter
    • Gauge
    • Summary
  • Labels
  • Const labels
  • Examples
    • Serve examples
      • Gauges
      • Summaries
      • How to use the examples
    • PushGateway examples
      • Gauges
      • How to use the examples
  • Tests
  • TODO
  • Author
  • License

Status

Under heavy development

Install

$ pip install prometheus

Why Python 3 and not python 2?

I think that everyone should start adopting the "new" Python version and let python2 be the old man that every one likes talking to but don't want live be with him.

And the only way doing this is by "forcing people" to use py3.

Also Maintaining code for one version is hard, imagine 2... error prone, slower updates...

So, don't use Python 2 and start using Python 3!

Usage

Serve data

from http.server import HTTPServer
from prometheus.exporter import PrometheusMetricHandler
from prometheus.registry import Registry

registry = Registry()

def handler(*args, **kwargs):
    PrometheusMetricHandler(registry, *args, **kwargs)

server = HTTPServer(('', 8888), handler)
server.serve_forever()

Push data (to pushgateway)

TODO

Metrics/Collectors

Counter

from prometheus.collectors import Counter

uploads_metric = Counter("file_uploads_total", "File total uploads.")
uploads_metric.inc({'type': "png", })

Gauge

from prometheus.collectors import Gauge

ram_metric = Gauge("memory_usage_bytes", "Memory usage in bytes.", {'host': host})
ram_metric.set({'type': "virtual", }, 100)

Summary

from prometheus.collectors import Summary

http_access =  Summary("http_access_time", "HTTP access time", {'time': 'ms'})

values = [3, 5.2, 13, 4]

for i in values:
    http_access.add({'time': '/static'}, i)

Labels

Labels define the multidimensional magic in prometheus. To add a metric to a collector you identify with a label for example we have this collector that stores the cosumed memory:

    ram_metric = Gauge("memory_usage_bytes", "Memory usage in bytes.")

And then we add our RAM user MBs:

    ram_metric.set({'type': "virtual", }, 100)

aplying mutidimensional capacity we can store in the same metric the memory consumed by the swap of our system too:

    ram_metric.set({'type': "swap", }, 100)

Const labels

When you create a collector you can put to than collector constant labels, these constant labels will apply to all the metrics gathered by that collector apart from the ones that we put. For example this example without const labels

    ram_metric = Gauge("memory_usage_bytes", "Memory usage in bytes.")
    ram_metric.set({'type': "virtual", 'host': host}, 100)
    ram_metric.set({'type': "swap", 'host': host}, 100)

is the same as this one with const labels:

    ram_metric = Gauge("memory_usage_bytes", "Memory usage in bytes.",  {'host': host})
    ram_metric.set({'type': "virtual", }, 100)
    ram_metric.set({'type': "swap", }, 100)

Examples

Serve examples

Gauges

• Memory and cpu usage (Requires psutil)
• Trigonometry samples

Summaries

• Disk write IO timing

How to use the examples

First some examples need requirements, install them:

pip install requirements_test.txt

Now run an example, for example timing_write_io_example.py

python ./examples/timing_write_io_example.py

All examples run on port 4444. You can point prometheus conf like this to point to one of the examples:

job: {
  name: "python-client-test"
  scrape_interval: "1s"
  target_group: {
    target: "http://xxx.xxx.xxx.xxx:4444/metrics"
  }
}

Or you can test the different formats available with curl:

Default (Text 0.0.4):

curl 'http://127.0.0.1:4444/metrics'

Text (0.0.4):

curl 'http://127.0.0.1:4444/metrics' -H 'Accept: text/plain; version=0.0.4'

Protobuf debug (0.0.4):

curl 'http://127.0.0.1:4444/metrics' -H 'Accept: application/vnd.google.protobuf; proto=io.prometheus.client.MetricFamily; encoding=text'

Protobuf (0.0.4):

curl 'http://127.0.0.1:4444/metrics' -H 'Accept: application/vnd.google.protobuf; proto=io.prometheus.client.MetricFamily; encoding=delimited'

PushGateway examples

Gauges

• input digits

How to use the examples

First you need to run a gateway, for example with docker:

docker run  --rm -p 9091:9091 prom/pushgateway

Now configure prometheus to grab the metrics from the push gateway example

job: {
  name: "pushgateway"
  scrape_interval: "1s"
  target_group: {
    target: "http://172.17.42.1:9091/metrics"
  }
}

Ready to launch the example:

python ./examples/input_example.py

As the serve explanation, you can debug de pushgateway serving data by accessing its URL (in the example: http://localhost:9091/metrics) with Curl

Tests

$ pip install -r requirements_test.txt
$ ./run_tests.sh

TODO

• Moaaaar examples
• implement handy utils

Author

Xabier (slok) Larrakoetxea

License

See License