All of the rev5s are decommissioned now. Let's get them out of buildbot-configs.

Fun fact, looks like bug 1209258 scheduled the MSE Video Playback tests on the OSX 10.10 rev5 machines back in the day and nobody ever got them running on the rev7s. As a result, we've had no OSX coverage since the rev5s were decommissioned roughly 10 months ago. I'll fix that while I'm at it.

This'll finish the job from bug 1226180 as well.

As noted, this also fixes what I'm betting was the unintentional disabling of the MSE tests on OSX 10.10.

Here's what's going to be getting scheduled again. Does this look right to you, Chris?

Comment on attachment 8811129 [details] [diff] [review] (buildbot-configs) remove the rev5 yosemite buildbot configs Review of attachment 8811129 [details] [diff] [review]: ----------------------------------------------------------------- My *only* concern is that those mediatests may suddenly start failing and confuse sheriffs (thinking unrelated patches broke it). Please be sure they (as well as media people) know that we're re-enabling them on OSX. ::: mozilla-tests/BuildSlaves.py.template @@ -11,5 @@ > "ubuntu64-asan_vm": "pass", > "ubuntu64-asan_vm_lnx_large": "pass", > "ubuntu64_hw": "pass", > "snowleopard": "pass", > - "yosemite": "pass", noteworthy, this template doesn't actually get used on "production" buildbot masters (this file there is populated via a copy in the puppet repo) ::: mozilla-tests/config.py @@ +2709,3 @@ > if slave_platform in branch['platforms'][platform]: > branch['platforms'][platform][slave_platform]['opt_unittest_suites'] += MEDIATESTS > branch['platforms'][platform][slave_platform]['debug_unittest_suites'] += MEDIATESTS we could just add these MEDIATESTS by default and get rid of this whole block (no gecko < 44 these days) --doesn't have to be this bug.

(In reply to Justin Wood (:Callek) from comment #6) > Comment on attachment 8811129 [details] [diff] [review] > (buildbot-configs) remove the rev5 yosemite buildbot configs > > Review of attachment 8811129 [details] [diff] [review]: > ----------------------------------------------------------------- > > My *only* concern is that those mediatests may suddenly start failing and > confuse sheriffs (thinking unrelated patches broke it). Please be sure they > (as well as media people) know that we're re-enabling them on OSX. This patch is part of my stack that includes the Win8 e10s changes that'll need a reconfig, so I intend to be around when that happens. I'll make sure they get hidden if there in fact do turn out to be problems. > ::: mozilla-tests/BuildSlaves.py.template > @@ -11,5 @@ > > "ubuntu64-asan_vm": "pass", > > "ubuntu64-asan_vm_lnx_large": "pass", > > "ubuntu64_hw": "pass", > > "snowleopard": "pass", > > - "yosemite": "pass", > > noteworthy, this template doesn't actually get used on "production" buildbot > masters (this file there is populated via a copy in the puppet repo) Should I drop this change or is it OK to leave in? > ::: mozilla-tests/config.py > we could just add these MEDIATESTS by default and get rid of this whole > block (no gecko < 44 these days) --doesn't have to be this bug. Yeah, I intend to do that as part of a bigger refactoring in bug 1276377. In general, I'd like to make more of the test suites run by default with ESR45 explicitly opting out where necessary. I think that'll end up being a lot less code and easier to follow.

OK to leave in

Comment on attachment 8811130 [details] builder diff Yes, I assume e10s media-tests will be run on these machines too.

Comment on attachment 8811132 [details] [diff] [review] (slave_health) remove yosemite_r5 and mtnlion_r5 references from slave_health https://hg.mozilla.org/build/slave_health/rev/b1bf0595c7797397ade13244d83baeac64811911

Comment on attachment 8811129 [details] [diff] [review] (buildbot-configs) remove the rev5 yosemite buildbot configs https://hg.mozilla.org/build/buildbot-configs/rev/3dda8f45e67447d150c2ab638e14f1856096b727

In production: https://hg.mozilla.org/build/buildbot-configs/rev/3dda8f45e674

t-yosemite-r5-0001 and t-yosemite-r5-0002 still exist, could someone followup with the appropriate decomission bug ? Now disabled in slavealloc.

n/m, I added a comment on bug 1226180.