1093897 - mozharness mach try builds can't do less than mozilla-release and older builds

Status: RESOLVED FIXED

(Release Engineering :: Applications: MozharnessCore, defect)

Description

[Jordan Lund (:jlund)]

my understanding is that try is currently able to do m-a and m-b (along with the default m-c).

mozharness mach builds (1055918) is not able to do m-a,b however we still want to enable mozharness mach on try tomorrow.

If things are all perfect, we merely need to uplift all the mach/build-system in-tree revs to m-a,b and things should 'just work'. Since we have hit so many blockers already on this rollout, I'd rather not have this bug block but instead be immediate follow up work. i.e. breaking m-a,b on try for 'a few days'

opening this bug for reference and discussion.

Comment 1

[Ryan VanderMeulen [:RyanVM]]

The next uplift is in just over 2 weeks and we'll want to be running Aurora-as-Beta Try pushes Really Soon Now. As long as "for a few days" is really that, I think we'll be OK. But this isn't a bug that can afford to drag out long.

Also, we'll want the various patches ported to the B2G release branches and esr31 as well.

Comment 2

[Michael Shal [:mshal]]

uplifted to aurora:
remote:   https://hg.mozilla.org/releases/mozilla-aurora/rev/43dcdc7f4626
remote:   https://hg.mozilla.org/releases/mozilla-aurora/rev/14de6cbff9c2
remote:   https://hg.mozilla.org/releases/mozilla-aurora/rev/f81cf71ec0cd
remote:   https://hg.mozilla.org/releases/mozilla-aurora/rev/0ebb0de0bb58

and beta:
remote:   https://hg.mozilla.org/releases/mozilla-beta/rev/8201c42832ef
remote:   https://hg.mozilla.org/releases/mozilla-beta/rev/6ab542eb236d
remote:   https://hg.mozilla.org/releases/mozilla-beta/rev/daabf4d8995f
remote:   https://hg.mozilla.org/releases/mozilla-beta/rev/8a6160e2ef98
remote:   https://hg.mozilla.org/releases/mozilla-beta/rev/5591e0a83c4d
remote:   https://hg.mozilla.org/releases/mozilla-beta/rev/72d8ba95b2db
remote:   https://hg.mozilla.org/releases/mozilla-beta/rev/ee2c3cfb4a7b
remote:   https://hg.mozilla.org/releases/mozilla-beta/rev/dc18ad2b4816

Comment 3

[Jordan Lund (:jlund)]

try *should* be able to do aurora and beta builds now. this was the primary concern.

as for release, esr31, and b2g release builds, we should follow up in this bug and keep blocking  	1055918 until that's resolved.

Comment 4

[Michael Shal [:mshal]]

Is there any reason to hold off on esr31, b2g30, and b2g32? Or should I just do them now?

And is it just those three? Or do I need mozilla-release or anything else as well?

Comment 5

[Jordan Lund (:jlund)]

(In reply to Michael Shal [:mshal] from comment #4)
> Is there any reason to hold off on esr31, b2g30, and b2g32? Or should I just
> do them now?
> 
> And is it just those three? Or do I need mozilla-release or anything else as
> well?

if it doesn't effect our current automation (buildbot based w/ 'make compile'), we should do it now. I don't how it would though since they should be independent.

IIUC mozilla-release is needed as well as we sometimes test mainline m-c on try.

Comment 6

[Ryan VanderMeulen [:RyanVM]]

Any time SGTM. I don't think we need to worry about m-r at this point given that mozilla-beta is going to be merging to it next week anyway.

Comment 7

[Michael Shal [:mshal]]

uplift to esr31: https://hg.mozilla.org/releases/mozilla-esr31/pushloghtml?changeset=6bfb385b357a

uplift to b2g32: https://hg.mozilla.org/releases/mozilla-b2g32_v2_0/pushloghtml?changeset=62294be0fc98

I also tried to uplift to b2g30, but the patches required a bit more modification to get them to apply. I also had some trouble with pymake when I pushed to try. In IRC, RyanVM mentioned that b2g30 is close enough to EOL that it's probably not worth the effort. So I'm not planning to do b2g30 unless someone makes a compelling case.

Comment 8

[Jordan Lund (:jlund)]

I am going to close this bug as we are in release week and bumping b2g30 anyway