The metrics page filters out iterations that ended up with EXCEPTION or RETRY, but it doesn't filter out ones that failed due to infrastructure misbehaving. For example, a failure to download built archive counts as if it provided testing results for this revision. This happens quickly, so we get an incorrect small result for the elapsed time. We should be more selective about which iterations we consider productive.
Created attachment 237262 [details] proposed fix
Comment on attachment 237262 [details] proposed fix View in context: https://bugs.webkit.org/attachment.cgi?id=237262&action=review > Tools/BuildSlaveSupport/build.webkit.org-config/public_html/dashboard/Scripts/BuildbotIteration.js:71 > +BuildbotIteration.ProductiveSteps = { Not an array? > Tools/BuildSlaveSupport/build.webkit.org-config/public_html/dashboard/Scripts/BuildbotIteration.js:295 > + var finishedAProductiveStep = false; s/A/Any/?
Committed <http://trac.webkit.org/r173027>. > Not an array? I do an "in" on it, which should be faster than an indexOf on an array.