metrique provides a simple python and rest API to support ETL workloads for extracting data from disperate sources, iteratively, rapidly and reproducibly, with transparent, historical serverside object persistence and tight clientside integration with popular python scientific computing libraries to faciliate creation and publication of a wide variety of analysis and reports, large and small.
Author: "Chris Ward" <cward@redhat.com> Sources: https://github.com/kejbaly2/metrique
The instructions given below assume fedora rpm package names:
# prerequisite *os* packages
sudo yum install python python-devel python-setuptools python-pip
sudo yum install git gcc gcc-c++ gcc-gfortran
sudo yum install freetype-devel libpng-devel # matplotlib deps
# metriqued - mongodb expected to be running; kerberos is optional
sudo yum install mongodb mongodb-server krb5-devel
# metriqued - nginx is optional
sudo yum install nginx
# additional python global dependencies, from pip
sudo pip install pip-accel # faster cached pip installed
# make sure our core package managers are up2date
sudo pip-accel install -U distribute setuptools
# our installation directory is always a python virtualenv
sudo pip-accel install virtualenv
# get the metrique sources
git clone https://github.com/kejbaly2/metrique.git
cd metrique
# deploy metrique master branch into a virtual environment
# including dependencies.
# NOTE this takes 10-15 minutes to compile everything from source!
./metrique -V ~/metrique.master deploy --ipython --pytest --docs --develop
# activate the virtual environment
source ~/virtenv-metrique/bin/activate
# optionally, start mongodb and metriqued
./metrique mongodb firstboot
./metrique mongodb start
./metrique metriqued firstboot
./metrique metriqued start
# launch ipython, connect to a metriqued instance and start mining!