Prometheus Pandas

Python library for querying Prometheus and accessing the results as Pandas data structures.

This is mostly intended for use in Jupyter notebooks. See Prometheus.ipynb for an example.

Example

Evaluate an instant query at a single point in time:

>>> from prometheus_pandas import query
>>>
>>> p = query.Prometheus('http://localhost:9090')
>>> series = p.query('node_cpu_seconds_total{mode="system"}', '2020-05-10T00:00:00Z')
>>> print(series)
node_cpu_seconds_total{cpu="0",instance="localhost:9100",job="node",mode="system"}    15706.47
node_cpu_seconds_total{cpu="1",instance="localhost:9100",job="node",mode="system"}    15133.25
node_cpu_seconds_total{cpu="2",instance="localhost:9100",job="node",mode="system"}    15095.59
node_cpu_seconds_total{cpu="3",instance="localhost:9100",job="node",mode="system"}    14649.20
dtype: float64

Evaluates an expression query over a time range:

>>> from prometheus_pandas import query
>>>
>>> p = query.Prometheus('http://localhost:9090')
>>> dataframe = p.query_range(
        'sum(rate(node_cpu_seconds_total{mode=~"system|user"}[1m])) by (mode)',
        '2020-10-05T00:00:00Z', '2020-10-05T06:00:00Z', '1h')
>>> print(dataframe)
dtype: float64
---
                     {mode="system"}  {mode="user"}
2020-10-05 00:00:00         0.022667       0.038222
2020-10-05 01:00:00         0.015333       0.036667
2020-10-05 02:00:00         0.028000       0.040667
2020-10-05 03:00:00         0.015111       0.034889
2020-10-05 04:00:00         0.015556       0.038000
2020-10-05 05:00:00         0.018444       0.040222
2020-10-05 06:00:00         0.018222       0.035111

Customizing the HTTP request:

>>> import requests
>>> from prometheus_pandas import query
>>>
>>> http = requests.Session()
>>> http.auth = ('user', 'pass')  # Basic authentication
>>> http.cert = '/path/client.cert'  # X.509 client certificate authentication
>>> http.verify = '/path/to/certfile'  # Custom certificate bundle
>>>
>>> p = query.Prometheus('http://localhost:9090', http)
>>> series = p.query('node_cpu_seconds_total{mode="system"}', '2020-10-05T00:00:00Z')
>>> print(series)
node_cpu_seconds_total{cpu="0",instance="localhost:9100",job="node",mode="system"}    3954.92
dtype: float64

Installation

Latest release via pip:

pip install prometheus-pandas [--user]

via Git:

git clone https://github.com/dcoles/prometheus-pandas.git; cd prometheus-pandas
python3 setup.py install [--user]

Licence

Licenced under the MIT License. See LICENSE for details.