Kim: just to be absolutely clear here, based on bug 1033507, we can decomm *all* tegras when Firefox 32 hits the release branch on September 2nd? Amy is looking for a date so we can start planning for the physical decomm process. I had a look at which branches tegras are currently enabled on: * mozilla-release * mozilla-b2g28_v1_3 * mozilla-b2g30_v1_4 The two b2g branches don't seem to actually generate any test jobs though. We could probably disable those now if we cared enough.
Aren't we going to need tegras for the theoretically-to-be-enabled Armv6 builds coming off esr31?
Yes, the builders will be disabled with bug 1033507 for 32. However, with bug 1042845 and bug 1036571 regarding enabling ESR support for Android 2.2, my assumption was that once this was online, tegras would be needed for ESR testing.
(In reply to Kim Moir [:kmoir] from comment #3) > Yes, the builders will be disabled with bug 1033507 for 32. However, with > bug 1042845 and bug 1036571 regarding enabling ESR support for Android 2.2, > my assumption was that once this was online, tegras would be needed for ESR > testing. We should verify with mobile, but the idea here was that Android 2.3 emulator testing was sufficient for supporting the EOL branch. We can still decomm the tegras with Firefox 32.
There are myriad code references and installations we'll need to clean up as part of the decomm. In no particular order: * buildbot-configs * buildbotcustom * tools * puppet * mozharness * slavealloc * slave health * slaveapi * buildapi * braindump * graphserver * tbpl * dev-master mobile staging instances * reclaim foopies
* treeherder-ui * treeherder-service * orangefactor * logparser * talos
Kim: do you have time to start writing patches for some of the key releng repos? I'm most concerned about buildbot-configs/buildbotcustom/tools/mozharness because I think there's possibility for substantial code removal. I can take a stab at the patches if you're too busy.
I can start looking at buildbot-configs/buildbotcustom/tools/mozharness to start. I love ripping out old code. It's like taking a sledgehammer to a wall.
What's the status here? Will this goal hit?
I removed the tegra code from slave health: https://hg.mozilla.org/build/slave_health/rev/00c8865e30e5
(In reply to Laura Thomson :laura from comment #9) > What's the status here? Will this goal hit? We've already hit the nominal goal in that we're no longer testing on tegras. It's just removing code and doing cleanup at this point.
Slavealloc databases (both production and staging) have been updated to remove tegras.