Collecting together the issues contributing to the 70%+ failure rate we're seeing at the moment.
I'd like to point out that the 70% fail rate is only on m-c and aurora based branches. The mozilla-b2g18 branch is pretty stable green from what I can see.
I'm removing dependencies for disabled tests which cause them not to impact TBPL/sheriffing any longer
Nearly all the webapi issues are likely the same problem, which can be summed up as: random tests fail after running the telephony tests. In bug 831830 I re-ordered the tests, so that Marionette unit tests are run before the webapi tests. Since this, there have been no failures in non-webapi tests. I will be closing several related bugs. I plan to file another bug to reorder to the webapi tests themselves, to see if we can get rid of some of the other intermittent failures.
WebAPI Tests are no longer run so am closing these as wontfix