Analytics/Archive/Stat1

This page is archived! Find up-to-date documentation at https://wikitech.wikimedia.org/wiki/Analytics

Background

This page documents the use cases for the stat1 server. Stat1 has two primary uses:

• Replacement of Bayes
• Development box for the Analytics/Reportcard project

Requested Services

incoming

• SSH access either directly or through fenari.
• Mediawiki should be installed and running, available to a general audience via HTTP.

outgoing

• Access to a mysql server
• Access to NFS served from dataset2 ?

software requirements for reportcard pipeline processing

• git
  • need to be able to clone and pull the reportcard repository there.
• pip
  • http://www.saltycrane.com/blog/2010/02/how-install-pip-ubuntu/
  • need this to install python packages
• virtualenv
  • http://pypi.python.org/pypi/virtualenv
  • Once this is installed, I can install any further python packages in my home directory.
• mysql client (and headers/libs)
  • Need this to install MySQL-python package (with pip/virtualenv)
• GeoIP C library
  • http://www.maxmind.com/app/c

Suggested Solution

• The db for the reportcard project will be put on db9, this is only intended for dev purposes, not production.
• Before it goes into production, we give better estimates on load

Relevant tickets

• reportcard requirements on stat1
• Stat1 should be able to access db9 mysql server
• move stat1 to an internal IP address
• Install mysql-server on stat1
• puppetize the statistics server setup (bayes, stat1)
• config changes stat1
• Add plotting libraries to stat1
• add ip->geo support to stats1