Skip to content

Latest commit

 

History

History
259 lines (160 loc) · 5.23 KB

README.rst

File metadata and controls

259 lines (160 loc) · 5.23 KB
Raw

python-libmetric

Python library for querying metrics from several time-series databases into Pandas DataFrames.

It support two types of metric queries, the first is instant metric, returning the value in precise moment in time. The second is the range metric, giving you the series of values for given time range and step.

Installation

Install the required dependencies on Debian based systems.

apt-get -y install librrd-dev libpython-dev

Install library from pip package.

pip install libmetric

Install library from source.

git clone https://github.com/cznewt/python-libmetric.git
cd python-libmetric
python setup.py install

Input Parameters

Parameters can be either set by environmental parameters or passed as command arguments.

For example passing the parameters as environmental parameters.

export LIBMETRIC_ENGINE='prometheus'
export LIBMETRIC_URL='https://metric01:9090'
export LIBMETRIC_QUERY='alertmanager_notifications_total'

export LIBMETRIC_START='2017-11-12T00:00:00Z'
export LIBMETRIC_END='2017-11-16T00:00:00Z'
export LIBMETRIC_STEP='3600s'

range_metric

And the example of passing parameters as command arguments.

range_metric --engine prometheus --url 'https://metric01:9090' --query '...'

Common Parameters

LIBMETRIC_ENGINE
Type of the endpoint to make query.
LIBMETRIC_URL
URL of the endpoint service.
LIBMETRIC_PARTITION
Data partition on target service endopoint.
LIBMETRIC_QUERY
Query to get the metric time-series or value.

Range Parameters

Parameters that apply only for the range meters.

LIBMETRIC_START
Time range start.
LIBMETRIC_END
Time range end.
LIBMETRIC_STEP
Query resolution step width.

Instant Parameters

Parameters that apply only for the intant meters.

LIBMETRIC_MOMENT
Single moment in time.

Alarm Parameters

Parameters that apply only for the all meters/alarms. Except the LIBMETRIC_AGGREGATION is applicable only for range meters.

LIBMETRIC_ALARM_THRESHOLD
Threshold for the alarms.
LIBMETRIC_ALARM_OPERATOR
Arithmetic operator for alarm evaluation. [gt, lt, gte, lte, eq]
LIBMETRIC_AGGREGATION
Aggregation function for the given time-series [min, max, sum, avg]

Supported Endpoints

The libmetric supports several major time-series databases to get the results in normalised way. The endpoints are queried thru HTTP API calls.

Graphite

Example configuration to query the Graphite server.

export LIBMETRIC_ENGINE='graphite'
export LIBMETRIC_URL='http://graphite.host:80'
export LIBMETRIC_QUERY='averageSeries(server.web*.load)'
...

InfluxDb

Example configuration to query the InfluxDb server.

export LIBMETRIC_ENGINE='influxdb'
export LIBMETRIC_URL='http://influxdb.host:8086'
export LIBMETRIC_USER='user'
export LIBMETRIC_PASSWORD='password'
export LIBMETRIC_PARTITION='prometheus'
export LIBMETRIC_QUERY='SELECT mean("value") FROM "alertmanager_notifications_total"'
...

Prometheus

Example configuration to query the Prometheus server.

export LIBMETRIC_ENGINE='prometheus'
export LIBMETRIC_URL='https://prometheus.host:9090'
export LIBMETRIC_QUERY='alertmanager_notifications_total'
...

Round-Robin Database

Example configuration to query the RRD file. The query is the consolidation function and the partition is the data set.

pip install python-rrdtool


export LIBMETRIC_ENGINE='rrd'
export LIBMETRIC_URL='file:///tmp/port.rrd'
export LIBMETRIC_PARTITION='INOCTETS'
export LIBMETRIC_QUERY='AVERAGE'
...

Usage from python

from libmetric.query_range import PrometheusRangeQuery

query = 'cpu{metric="load"}'

url = 'http://localhost:9090'

data = {
    'queries': [query],
    'url': url,
    'step': step,
    'start': start,
    'end': end
}

query = PrometheusRangeQuery(**data)

print query.get()

Alarm Options

Following lists show allowed values for alarm functions, the alarm arithmetic operators and aggregation function for range meters.

Supported Alarm Operators

gt
Greater than >.
gte
Greater or equal than >=.
lt
Lower than <.
lte
Lower or equal than <=.
eq
Equals to ==.

Supported Time-series Aggregations

avg
Arithmetic average of the series values.
min
Use the minimal value from series.
max
Use the maximal value from series.
sum
Sum the values together.

More Information