navdoon 0.3.0

Powerful Statsd server, made easy

Powerful Statsd server, made easy.

Navdoon is a portable Statsd server with useful features to make it easy to use, extend and integrate.

Features

• Portable with few dependencies, easy to install on most platforms

• Support TCP, UDP

• Receive metrics from multiple addresses

• Flush to multiple Graphite backends

• Flush to files (in Graphite and CSV format)

• Easy to integrate with custom programs

• Ability to reload the server without losing the metrics

Details

• Navdoon uses collectors to receive Statsd metrics and recieves metrics over UDP (–collect-udp) and TCP (–collect-tcp), and accepts multiple collectors.

• The server saves/sends (flushes) the accumulated metrics every often (–flus-interval) to a persistent storage. Carbon (from Graphite project) is a very common backend for Statsd servers. Navdoon accepts multiple Graphite addresses (–flush-graphite) so it can flush to multiple backends (all share the same interval). Metrics can be flushed to standard output (–flush-stdout) to pipe to another program, so it’s easy to integrate with any custom backend.

• Logging can be helpful or can be wasteful, depending on the deployment and the usage of the application. Navdoon provides detailed configuration on logging, so you can chose what will be logged (–log-level) and how to log, send logs to syslog (–log-syslog), to a file (–log-file) or standard error (–log-stderr) to be piped to another program.

• While not claiming to be the fastest, good performance is considered in the design. Navdoon uses threads for each collector and flush backend. Future versions will offer improved performance as it was not a priority for the first releases.

• Server supports reloading (on receiving SIGHUP), keeping current state of the metrics and last flush time. So it’s possible to change collectors, flush destinations, logging, etc. on the configuration file while the server is running, and then on sending a SIGHUP the server picks the new configuration.

Releases

• Latest released version is 0.3.0 (released on 2017-02-11)

• Version 0.2.0 was released on 2016-10-10

See the CHANGELOG for more information about features provided by each release.

Install

Navdoon is distributed as a Docker image, and a Python package.

PyPi

Navdoon can be installed from pypi using pip.

pip install navdoon

Source

You can install from the source by running the setup.py script provided.

python setup.py install

Note

If you’re installing Navdoon to a system path, you might need to run the installation with sudo or under a privileged user.

Requirements

Navdoon is written in Python, so running from source or installing it as a package, requires a Python runtime (version 2.7+, latest versions of Python 3 is recommended).

The statsdmetrics Python module is the only dependency to run Navdoon. However these Python modules are recommended on development/test environment:

• distutilazy (>=0.4.2): helpful commands added to setup.py to run tests and clean temp files

• typing (>=3.5.0): standard type annotations for Python

• coverage: create test coverage reports

Docker

Different Docker image tags are available, providing different runtimes.

The default tag is based on CPython alpine images (trying to use latest versions of Python, currently 3.6) so the image size should be minimum.

docker pull farzadghanei/navdoon
# or a specific version like farzadghanei/navdoon:0.3.0

Another tag is available to run Navdoon on PyPy (currently version 2).

docker pull farzadghanei/navdoon:0.3.0-pypy2

Running from source

Before running from source, a few dependencies should be installed. Using a virtual environment is suggested. (In this example we create a virtual environment in the project source path, but you may chose a custom path like ~/.venvs/navdoon-py3)

git clone https://github.com/farzadghanei/navdoon.git && cd navdoon
python3 -m venv .navdoon-venv-py3 && source ./.navdoon-venv-py3/bin/activate
pip install -r requirements.txt && python3 bin/navdoon_src

Note

Python 3.3+ standard library comes with venv module. For older versions you can use virtualenv.

Or you may skip installing and sourcing the virtual environment and install the (few) dependencies on your system.

License

Navdoon is released under the terms of the Apache 2.0 license.