Closed Bug 818378 Opened 12 years ago Closed 12 years ago

releng changes needed for mozilla-b2g18 branch

Categories

(Release Engineering :: General, defect)

defect
Not set
normal

Tracking

(Not tracked)

RESOLVED FIXED

People

(Reporter: hwine, Assigned: hwine)

References

Details

(Whiteboard: [re-b2g])

Attachments

(2 files, 2 obsolete files)

changes to allow the mozilla-b2g18 branch to be used for nightly and CI b2g builds
Attached patch build/buildbot-configs changes (obsolete) — Splinter Review
Not fully tested in staging yet, but wanted to get some initial feedback - bb will start, and CI builds will get scheduled (lack of mock slaves is a holdup)
Attachment #688623 - Flags: feedback?(rail)
Attachment #688623 - Flags: feedback?(bhearsum)
Attached patch build/tools changes (obsolete) — Splinter Review
looking for initial feedback - build "works" on staging (no mock builders handy this evening) - CI changes are scheduled
Attachment #688624 - Flags: feedback?(rail)
Attachment #688624 - Flags: feedback?(bhearsum)
Comment on attachment 688623 [details] [diff] [review]
build/buildbot-configs changes

typo - these are changes for build/buildbot-configs
Attachment #688623 - Attachment description: build/tools changes → build/buildbot-configs changes
Comment on attachment 688623 [details] [diff] [review]
build/buildbot-configs changes

Review of attachment 688623 [details] [diff] [review]:
-----------------------------------------------------------------

You'll definitely need to touch b2g_config.py in both mozilla/ and mozilla-tests/. This patch just made me realize that we may _only_ want this branch there, too. Can you Alex if we want Desktop and Android builds on this branch? Holding off looking at anything else until we get that answered.

Should've realized this when we chatted last week, sorry :(
Attachment #688623 - Flags: feedback?(bhearsum)
Comment on attachment 688624 [details] [diff] [review]
build/tools changes

Review of attachment 688624 [details] [diff] [review]:
-----------------------------------------------------------------

production-branches.json looks fine. The mozconfig whilelist changes are only part of what you need. But if this branch ends up b2g-only, we don't need this branch in that file at all.
Comment on attachment 688623 [details] [diff] [review]
build/buildbot-configs changes

Review of attachment 688623 [details] [diff] [review]:
-----------------------------------------------------------------

In overall the patch looks good. I haven't checked if there are missing things though. I'd suggest to dump branch config for this branch and diff it against beta (and maybe esr17). Repeat this operation for config.py, b2g_config.py and thunderbird_config.py to be sure.

::: mozilla/b2g_config.py
@@ +778,5 @@
>  BRANCHES['mozilla-beta']['platforms']['unagi']['enable_nightly'] = True
>  BRANCHES['mozilla-beta']['platforms']['unagi']['nightly_signing_servers'] = 'nightly-signing'
>  BRANCHES['mozilla-beta']['platforms']['unagi']['mozharness_config']['extra_args'] = ['--target', 'unagi', '--config', 'b2g/releng-beta.py']
> +# bug https://bugzil.la/815185#c1 START - delete unagi_stable lines when
> +# switch to mozilla-b2g18

Hmmmm, not sure if unagi_stable should be removed. The nightly build uses "nightly" built-in channel for MAR files, while unagi_stable nightly uses "beta" (will be today). There would be no other way to build "beta" channel MAR if unagi_stable is removed.

::: mozilla/config.py
@@ +1033,5 @@
> +            'macosx64': {},
> +            'linux-debug': {},
> +            'linux64-debug': {},
> +            'macosx64-debug': {},
> +            'win32-debug': {},

BTW, why not all platforms?

@@ +1489,5 @@
> +BRANCHES['mozilla-b2g18']['l10nNightlyUpdate'] = False
> +BRANCHES['mozilla-b2g18']['l10n_platforms'] = ['linux', 'linux64', 'win32',
> +                                               'macosx64']
> +BRANCHES['mozilla-b2g18']['l10nDatedDirs'] = True
> +BRANCHES['mozilla-b2g18']['l10n_tree'] = 'fxesr17'

l10n_tree looks wrong

@@ +1502,5 @@
> +BRANCHES['mozilla-b2g18']['enable_blocklist_update'] = False
> +BRANCHES['mozilla-b2g18']['blocklist_update_on_closed_tree'] = False
> +BRANCHES['mozilla-b2g18']['enable_valgrind'] = False
> +BRANCHES['mozilla-b2g18']['enabled_products'] = ['firefox']
> +# mock disabled block start

I believe you can remove the whole "mock disabled" block and use default platform values.
Attachment #688623 - Flags: feedback?(rail) → feedback+
Comment on attachment 688624 [details] [diff] [review]
build/tools changes

Review of attachment 688624 [details] [diff] [review]:
-----------------------------------------------------------------

I believe you don't need to update the white list, since we're not going to use that branch in release automation.
Attachment #688624 - Flags: feedback?(rail) → feedback+
Comment on attachment 688623 [details] [diff] [review]
build/buildbot-configs changes

We clarified about the products/platforms in e-mail, here's what Alex said:
We don't need Fennec, but I think it makes sense to build Firefox given the fact that we may want to easily attempt reproducing security issues on the branch, and it's good data if we have build failures on those platforms.

--

Hal, what this means for buildbot-configs is the following:
* You want all of the desktop platforms listed in the config.py platforms list (linux, linux64, macosx64, win32, linux-debug, linux64-debug, macosx64-debug, and win32-debug for mozilla/config.py -- macosx, macosx64, win32, linux, linux64 for mozilla-tests/config.py)
* For b2g platforms, this patch looks correct already.
Attachment #688623 - Attachment is obsolete: true
Comment on attachment 688624 [details] [diff] [review]
build/tools changes

Review of attachment 688624 [details] [diff] [review]:
-----------------------------------------------------------------

Agreed - r=me with the mozconfig whitelist changes dropped.
Attachment #688624 - Flags: feedback?(bhearsum) → feedback+
Attachment #688624 - Attachment is obsolete: true
Attachment #689759 - Flags: review?(bhearsum)
appears to run okay in staging - diff of dumps not yet done
Attachment #689762 - Flags: review?(rail)
Attachment #689759 - Flags: review?(bhearsum) → review+
Comment on attachment 689762 [details] [diff] [review]
build/buildbot-configs changes

Review of attachment 689762 [details] [diff] [review]:
-----------------------------------------------------------------

A nit.

::: mozilla-tests/config.py
@@ +1028,5 @@
> +######### mozilla-b2g18
> +BRANCHES['mozilla-b2g18']['release_tests'] = 5
> +BRANCHES['mozilla-b2g18']['repo_path'] = "releases/mozilla-b2g18"
> +BRANCHES['mozilla-b2g18']['platforms']['linux']['enable_mobile_unittests'] = True
> +#BRANCHES['mozilla-b2g18']['pgo_strategy'] = 'per-checkin'

I don't think that you need pgo_strategy setting since neither of listed platforms support/use it. However, it would be better to double check where/how this variable used.

Otherwise, lgtm
Attachment #689762 - Flags: review?(rail) → review+
Comment on attachment 689762 [details] [diff] [review]
build/buildbot-configs changes

checked in with 'pgo_strategy' line removed (set to None above, which is valid)

default:
http://hg.mozilla.org/build/buildbot-configs/rev/86aaf7fe078f
Attachment #689762 - Flags: checked-in+
Status: NEW → RESOLVED
Closed: 12 years ago
Resolution: --- → FIXED
in production
Product: mozilla.org → Release Engineering
Component: General Automation → General
You need to log in before you can comment on or make changes to this bug.

Attachment

General

Created:
Updated:
Size: