wavefront-pyformance 1.0.0

Wavefront Pyformance Library

wavefront-pyformance

This is a plugin for pyformance which adds Wavefront reporters (via proxy or direct ingestion) and an abstraction that supports tagging at the host level. It also includes support for Wavefront delta counters.

Requirements

Python 2.7+ and Python 3.x are supported.

pip install wavefront-pyformance

Usage

Wavefront Reporter

The Wavefront Reporters support tagging at the host level. Tags passed to a reporter will be applied to every metric before being sent to Wavefront.

Create Wavefront Reporter

You can create a WavefrontProxyReporter or WavefrontDirectReporter as follows:

import pyformance
from wavefront_pyformance import wavefront_reporter

reg = pyformance.MetricsRegistry()

# report metrics to a Wavefront proxy every 10s
wf_proxy_reporter = wavefront_reporter.WavefrontProxyReporter(
    host=host, port=2878, registry=reg,
    source='wavefront-pyformance-example',
    tags={'key1': 'val1', 'key2': 'val2'},
    prefix='python.proxy.',
    reporting_interval=10)
wf_proxy_reporter.start()

# report metrics directly to a Wavefront server every 10s
wf_direct_reporter = wavefront_reporter.WavefrontDirectReporter(
    server=server, token=token, registry=reg,
    source='wavefront-pyformance-exmaple',
    tags={'key1': 'val1', 'key2': 'val2'},
    prefix='python.direct.',
    reporting_interval=10)
wf_direct_reporter.start()

Flush and stop Wavefront Reporter

After create Wavefront Reporter, start() will make the reporter automatically reporting every reporting_interval seconds. Besides that, you can also call report_now() to perform reporting immediately. report_now() should also be called before you stop the reporter as follows:

wf_reporter.report_now()
wf_reporter.stop()

Delta Counter

To create a Wavefront delta counter:

import pyformance
from wavefront_pyformance import delta

reg = pyformance.MetricsRegistry()
d_0 = delta.delta_counter(reg, 'requests_delta')
d_0.inc(10)

Note: Having the same metric name for any two types of metrics will result in only one time series at the server and thus cause collisions. In general, all metric names should be different. In case you have metrics that you want to track as both a Counter and Delta Counter, consider adding a relevant suffix to one of the metrics to differentiate one metric name from another.

Wavefront Histogram

To create a Wavefront Histogram:

import pyformance
from wavefront_pyformance import wavefront_histogram

reg = pyformance.MetricsRegistry()
h_0 = wavefront_histogram.wavefront_histogram(reg, 'requests_duration')
h_0.add(10)