On Thu, Jun 26, 2014 at 11:37:54PM -0400, Brian Proffitt wrote:
Downloads of the Engine RPM file, it was determined, would be
indicative of actual oVirt installs. The allinone RPMs were not analyzed at this time,
since installing this RPM is a choice made during the installation process itself.
Tracking the Engine Reports and dwh RPMs was done to determine the popularity of these two
tools and to ensure their numbers were comparable.
To count downloads, data was gathered each month that listed the total downloads of every
file, a number derived from the total hits each file had, minus any 206 hits, which
indicated incomplete downloads. Key files for Engine, Engine Reports, and Engine dwh were
identified and the data filtered to include counts for each one of the files, in whatever
version released.
Tracking of Live and Node RPMs was already set up within the awstats reporting, and was
taken directly from awstats in each month.
If there is any part of this methodology that is in error, feedback is very much
appreciated.
You should know that since we now have a mirrorlist (which in terms of
hit count is already in the top 3) the number could be much lower than
the actual RPM downloads. I'd be interested how much traffic is served
by mirrors.