Despite timeout limit being reached. For instance, this is the only instance of fx-team mochitest-1 opt windows xp that has run today https://secure.pub.build.mozilla.org/buildapi/self-serve/fx-team/build/73687378 RyanVM|sheriffduty kmoir|buildduty: does SETA know about per-product builds? RyanVM|sheriffduty i.e. if build #10 is an android-only push, did I just get hosed out of WinXP tests for another 10 pushes? kmoir|buildduty RyanVM|sheriffduty: If only android tests are scheduled, the scheduling skipping will only be applied to that platform RyanVM|sheriffduty so in theory I'll get them on the next push? kmoir|buildduty that is my understanding RyanVM|sheriffduty why hasn't WinXP run a full set of tests since 4am PT today? https://treeherder.mozilla.org/#/jobs?repo=fx-team&fromchange=e4738a23d0c4&filter-searchStr=Windows%20XP%20opt%20Mochitest%20Mochitest%20M%283%29 RyanVM|sheriffduty better shown by https://treeherder.mozilla.org/#/jobs?repo=fx-team&fromchange=e4738a23d0c4&filter-searchStr=windows%20xp%20opt%20mochitest
Created attachment 8715009 [details] [diff] [review] fix idle timeout
I think the problem was that we were comparing the age of the most recent request instead of the age of the oldest request. This meant that for the idle timeout to work, the time between any two builds had to be longer than the timeout period. Anything shorter would effectively reset the timeout.
Comment on attachment 8715009 [details] [diff] [review] fix idle timeout I landed it http://hg.mozilla.org/build/buildbotcustom/rev/12ca89b8a20b so could reconfig with bug 1243877 at the same time
in production, thanks for fixing this!