Closed Bug 1403040 Opened 2 years ago Closed 2 years ago

spin `maple` back up to replace `date`

Categories

(Release Engineering :: General, enhancement)

enhancement
Not set

Tracking

(Not tracked)

RESOLVED FIXED

People

(Reporter: aki, Assigned: aki)

References

(Depends on 1 open bug)

Details

Attachments

(8 files)

18:57 <•catlee> can we rename date?
19:03 <•aki> haha
19:03 <•aki> we can spin up a new one and retire date
19:04 <•aki> it'll take a bit of effort, but then it'll save future confusion i think
19:06 <rail> "date" is a terrible name :)
19:07 <•aki> hahaha
19:10 <•aki> let's paint it `dogwood`. any other tree names?
19:11 — rail googles for "date in Persian" because they are the tastiest
19:12 <rail> using خرما would be a good UTF-8 stress on our systems :D
19:12 <•aki> haha
19:13 <rail> we can use "hurma" per https://en.wiktionary.org/wiki/%D8%AE%D8%B1%D9%85%D8%A7 :P
19:13 <•aki> don't you mean hurmâ ?
19:15 <•aki> i think we need a devsvcs bug, a tools patch, a scriptworker patch, a buildbot-configs patch, a new-branch patch
19:15 <rail> yes, but I don't think â is an easy thing to digest :)
19:15 <•aki> not sure if we need to create the branch in balrog manually?
19:15 <•aki> or if it autocreates new branches if they submit properly
19:16 <•aki> mozilla-taskcluster and treeherder
19:16 <rail> and treestatus

So we'll need all of those things, but first we need a name.
Summary: new project branch to replace `date` → spin `maple` back up to replace `date`
Depends on: 1403249
We'll need roles and hooks set up as well.
Assignee: nobody → aki
In braindump/buildbot-related/dump_master_json.py, we use multiprocessing. In our tools queuedir.py, we use the `if not os.path.exists(path): ...` construct, which is prone to race conditions in non-synchronous code.

I'm not sure if we want this patch, but it should help us avoid race condition bustage in our buildbot-configs tests.
Attachment #8913767 - Flags: review?(rail)
Attachment #8913792 - Flags: review?(rail)
Attachment #8913772 - Flags: review?(rail)
Attachment #8913776 - Flags: review?(rail)
Adding maple:

_ x devsvcs bug - bug 1403249
_ x a tools patch - attached
_ x a scriptworker patch - attached
_ x a buildbot-configs patch - attached
x x mozilla-taskcluster - already supports maple
_ x treeherder - attached
x x treestatus - added `maple` branch
_ _ a new-branch patch - will land when everything else is set up

Removing date:

_ _ devsvcs bug
_ x tools patch - attached
_ x scriptworker patch - attached
_ x buildbot-configs patch - attached
_ _ mozilla-taskcluster (not sure if we need this?)
_ _ treeherder
_ _ treestatus
Attachment #8913767 - Flags: review?(rail) → review+
Comment on attachment 8913772 [details] [diff] [review]
[tools] fix queuedir race condition bustage

hehe :)
A nit. Should we use os.path.isdir() instead of os.path.exists()?
Attachment #8913772 - Flags: review?(rail) → review+
Attachment #8913776 - Attachment is patch: true
Attachment #8913776 - Attachment mime type: text/x-github-pull-request → text/plain
Attachment #8913776 - Flags: review?(rail) → review+
Comment on attachment 8913792 [details] [diff] [review]
[buildbot-configs] replace date with maple

lgtm!
Attachment #8913792 - Flags: review?(rail) → review+
Attachment #8913818 - Flags: review?(rail) → review+
Adding maple:

_ x devsvcs bug - bug 1403249 - needs level 3 switch
x x a tools patch - attached
x x a scriptworker patch - attached
x x a buildbot-configs patch - attached
x x mozilla-taskcluster - already supports maple
x x treeherder - attached
x x treestatus - added `maple` branch
_ _ a new-branch patch - will land when everything else is set up

Removing date:

_ _ devsvcs bug
x x tools patch - attached
x x scriptworker patch - attached
x x buildbot-configs patch - attached
_ _ mozilla-taskcluster (not sure if we need this?)
_ _ treeherder
_ _ treestatus
Attachment #8914475 - Flags: review?(bugspam.Callek)
Comment on attachment 8914475 [details] [diff] [review]
maple-staging.patch

We don't want to land this patch anywhere but maple; r? -> f?
Attachment #8914475 - Flags: review?(bugspam.Callek) → feedback?(bugspam.Callek)
Comment on attachment 8914475 [details] [diff] [review]
maple-staging.patch

Review of attachment 8914475 [details] [diff] [review]:
-----------------------------------------------------------------

::: browser/branding/nightly/branding.nsi
@@ +14,5 @@
>  !define URLInfoAbout          "https://www.mozilla.org"
>  !define HelpLink              "https://support.mozilla.org"
>  
> +!define URLStubDownload32 "http://download.mozilla.org/?os=win&lang=${AB_CD}&product=firefox-maple-nightly-latest"
> +!define URLStubDownload64 "http://download.mozilla.org/?os=win64&lang=${AB_CD}&product=firefox-maple-nightly-latest"

it was mentioned to me that Stub shouldn't point elsewhere even on project branches, but I have no skin in that game and can easily assume we want this to point at "-maple-"

::: mobile/android/branding/nightly/configure.sh
@@ +6,5 @@
>  MOZ_UPDATER=1
>  MOZ_ANDROID_ANR_REPORTER=1
>  MOZ_ANDROID_SHARED_ID=org.mozilla.fennec.sharedID
>  MOZ_ANDROID_GCM_SENDERID=965234145045
> +MOZ_MMA_GCM_SENDERID=242693410970

I don't see the change on this line (10), EOL/EOF change?
Attachment #8914475 - Flags: feedback?(bugspam.Callek) → feedback+
Depends on: 1405157
I landed:

These patches should not be merged to central!
==============================================
- the above maple staging patch https://hg.mozilla.org/projects/maple/rev/023bf2fe47c5
- l10n-bumper, until it merges autoland -> m-c -> date. I reduced the desktop locales to ['de', 'zh-TW']. https://hg.mozilla.org/projects/maple/rev/35e2e062a2eb
- by default, use mozilla_beta_tasks as the target_tasks_method on-push https://hg.mozilla.org/projects/maple/rev/0262f2d8bac9

These patches are intended for central, once they bake and get review
=====================================================================
- bug 1397552 - latest relpro action task patch - https://hg.mozilla.org/projects/maple/rev/8df8bbd22a9e
- bug 1397774 - allow for relpro target_tasks_method (adds nightly_l10n to desktop relpro) https://hg.mozilla.org/projects/maple/rev/0b9f28299e8a
- bug 1397773 - add funsize-balrog-submitter and beet-mover gecko-image tasks (we can prevent these from optimizing out via `do_not_optimize` parameter) https://hg.mozilla.org/projects/maple/rev/bcfcd2ce7672

Treeherder deployed with maple enabled.
No decision task launched on push; I suspect that's due to bug 1405157.

My current plan is to add fennec bbb releasetasks to the fennec candidates target task, as a smaller problem set for in-tree relpro.
Adding maple:

x x devsvcs bug - bug 1403249 - needs level 3 switch
x x a tools patch - attached
x x a scriptworker patch - attached
x x a buildbot-configs patch - attached
x x mozilla-taskcluster - already supports maple
x x treeherder - attached
x x treestatus - added `maple` branch
x x a new-branch patch - will land when everything else is set up
_ _ tc-admin bug 1405157
_ _ hooks + roles

Removing date:

_ _ devsvcs bug
x x tools patch - attached
x x scriptworker patch - attached
x x buildbot-configs patch - attached
_ _ mozilla-taskcluster (not sure if we need this?)
_ _ treeherder
_ _ treestatus
_ _ tc-admin patch to kill date
_ _ remove date hooks + roles
Maple is usable!!
I'm not in a hurry to completely remove date just yet. If someone else has a strong opinion here, I'm happy to table date cleanup or push it along; otherwise we can get back to it when it's clear we don't need it anymore.

Adding maple:

x x devsvcs bug - bug 1403249 - needs level 3 switch
x x a tools patch - attached
x x a scriptworker patch - attached
x x a buildbot-configs patch - attached
x x mozilla-taskcluster - already supports maple
x x treeherder - attached
x x treestatus - added `maple` branch
x x a new-branch patch - will land when everything else is set up
x x tc-admin bug 1405157
_ _ hooks + roles - do we need these? We can possibly use the relpromo action without needing hooks?

Removing date:

_ _ devsvcs bug
x x tools patch - attached
x x scriptworker patch - attached
x x buildbot-configs patch - attached
_ _ mozilla-taskcluster (not sure if we need this?)
_ _ treeherder
_ _ treestatus
_ _ tc-admin patch to kill date
_ _ remove date hooks + roles
Depends on: 1407050
Depends on: 1407052
- I think we have sufficient hooks and roles, with the action tasks. If we need more, we can add more.
- I think mozilla-taskcluster gets its cues from other repos, e.g. treeherder.

Afaict, we're done here once the two open dep bugs are fixed and the treeherder patch lands.

Adding maple:

x x devsvcs bug - bug 1403249 - needs level 3 switch
x x a tools patch - attached
x x a scriptworker patch - attached
x x a buildbot-configs patch - attached
x x mozilla-taskcluster - already supports maple
x x treeherder - attached
x x treestatus - added `maple` branch
x x a new-branch patch - will land when everything else is set up
x x tc-admin bug 1405157
n/a hooks + roles - do we need these?

Removing date:

_ x devsvcs bug
x x tools patch - attached
x x scriptworker patch - attached
x x buildbot-configs patch - attached
n/a mozilla-taskcluster (not sure if we need this?)
_ x treeherder
x x treestatus
_ x tc-admin bug to kill date
x x remove date hooks + roles
I think we're done other than the devsvcs bug 1407050.
Status: NEW → RESOLVED
Closed: 2 years ago
Resolution: --- → FIXED
Component: General Automation → General
You need to log in before you can comment on or make changes to this bug.