Last modified: 2014-02-26 12:53:31 UTC
Given the amount of testing happening from WMF IP addresses we should exclude/flag these events so as not to pollute the data.
To note it from our discussion in IRC: The only case where we potentially _do_ want WMF IP events collected is on test and test2. So if it's not a pain, excluding all but those would be one of acceptable solutions AFAIK.
The potential skews from bugs that we're not able to diagnose as a result of being in a special-cased IP block seems larger than the skew from having our test events contaminate the data, so I don't want to filter out WMF-generated events altogether. Flagging them should be doable. We need to find out our IP range. Steven, Dario: can one of you check? I think James F. blocked the IP range from anon-editing Wikipedia, so if you look at current blocks you should be able to get the correct range from there.
Anon editing is definitely not blocked: http://en.wikipedia.org/w/index.php?title=User%3ADarTar%2FSandBox&diff=531353488&oldid=529077420
(In reply to comment #2) > The potential skews from bugs that we're not able to diagnose as a result of > being in a special-cased IP block seems larger than the skew from having our > test events contaminate the data, so I don't want to filter out WMF-generated > events altogether. Flagging them should be doable. We need to find out our IP > range. Steven, Dario: can one of you check? I think James F. blocked the IP > range from anon-editing Wikipedia, so if you look at current blocks you > should > be able to get the correct range from there. He only blocked on MediaWiki.org, IIRC.
(In reply to comment #4) > He only blocked on MediaWiki.org, IIRC. Are you able to see the ranges? If so, could you paste them here or (if they are sensitive) e-mail them to me?
I think the solution to this might simply be to provide a good opt-out mechanism.
[moving from MediaWiki extensions to Analytics product - see bug 61946]
