Closed Bug 1029851 Opened 5 years ago Closed 5 years ago

Tracking bug for 02-sep-2014 migration work

Categories

(Release Engineering :: General, defect)

defect
Not set

Tracking

(Not tracked)

RESOLVED FIXED

People

(Reporter: aki, Assigned: rail)

References

Details

Attachments

(16 files, 6 obsolete files)

898 bytes, patch
hwine
: review+
Details | Diff | Splinter Review
4.76 KB, patch
hwine
: review+
Details | Diff | Splinter Review
1.51 KB, patch
hwine
: review+
Details | Diff | Splinter Review
727 bytes, patch
hwine
: review+
Details | Diff | Splinter Review
10.29 KB, patch
hwine
: review+
Details | Diff | Splinter Review
922 bytes, patch
hwine
: review+
Details | Diff | Splinter Review
537 bytes, patch
hwine
: review+
Details | Diff | Splinter Review
892 bytes, patch
hwine
: review+
Details | Diff | Splinter Review
868 bytes, patch
hwine
: review+
Details | Diff | Splinter Review
1.38 KB, patch
hwine
: review+
Details | Diff | Splinter Review
672 bytes, patch
hwine
: review+
Details | Diff | Splinter Review
1.08 KB, patch
hwine
: review+
Details | Diff | Splinter Review
10.00 KB, patch
rail
: review+
Details | Diff | Splinter Review
9.96 KB, patch
rail
: review+
Details | Diff | Splinter Review
1.10 KB, patch
standard8
: review+
Details | Diff | Splinter Review
611 bytes, patch
rail
: review+
pmoore
: checked-in+
Details | Diff | Splinter Review
No description provided.
Depends on: 1029860
Depends on: 1022907
Depends on: 1038421
If there are Fennec 31.0b10 tag/relbranch conflicts during the Sept 1 mozilla-beta -> mozilla-release migration, see bug 1038421 for clues.
Assignee: nobody → rail
Depends on: 1054317
To be landed before we start the RC build.
Attachment #8473706 - Flags: review?(hwine)
Comment on attachment 8473706 [details] [diff] [review]
mozilla-release-bump.diff

Review of attachment 8473706 [details] [diff] [review]:
-----------------------------------------------------------------

lgtm! 6 weeks sure flies by
Attachment #8473706 - Flags: review?(hwine) → review+
Depends on: 1056089
Comment on attachment 8474648 [details] [diff] [review]
[legacy vcs sync] start syncing aurora to v2.1

Review of attachment 8474648 [details] [diff] [review]:
-----------------------------------------------------------------

lgtm
Attachment #8474648 - Flags: review?(hwine) → review+
Comment on attachment 8474652 [details] [diff] [review]
[new vcs sync]: start syncing aurora to v2.1

Review of attachment 8474652 [details] [diff] [review]:
-----------------------------------------------------------------

lgtm
Attachment #8474652 - Flags: review?(hwine) → review+
Comment on attachment 8474664 [details] [diff] [review]
Pause Gecko l10n vcs-sync

Review of attachment 8474664 [details] [diff] [review]:
-----------------------------------------------------------------

lgtm
Attachment #8474664 - Flags: review?(hwine) → review+
Comment on attachment 8474668 [details] [diff] [review]
add new Gaia and l10n to vcs-sync

Review of attachment 8474668 [details] [diff] [review]:
-----------------------------------------------------------------

lgtm
Attachment #8474668 - Flags: review?(hwine) → review+
Rail: you also need the legacy vcs-sync patches for the l10n changes. That's the current "live" system.
Flags: needinfo?(rail)
(In reply to Hal Wine [:hwine] (use needinfo) from comment #12)
> Rail: you also need the legacy vcs-sync patches for the l10n changes. That's
> the current "live" system.

Thanks for the reminder. That part has always been a bit unclear for me. I'll prep the patches today/tomorrow.
Flags: needinfo?(rail)
Comment on attachment 8478286 [details] [diff] [review]
merge-day-buildbot-configs.diff

Review of attachment 8478286 [details] [diff] [review]:
-----------------------------------------------------------------

lgtm - nice catch on nexus eng build
Attachment #8478286 - Flags: review?(hwine) → review+
In production with reconfig on 2014-08-25 08:49 PT
Attachment #8474648 - Attachment filename: repo-sync-tools.diff → [legacy vcs sync] start syncing aurora to v2.1
Attachment #8474648 - Attachment description: repo-sync-tools.diff → [legacy vcs sync] start syncing aurora to v2.1
Attachment #8474648 - Attachment filename: [legacy vcs sync] start syncing aurora to v2.1 → repo-sync-tools.diff
Attachment #8474652 - Attachment description: mozharness_sync_aurora_to_v2.1.diff → [new vcs sync]: start syncing aurora to v2.1
Attachment #8474664 - Attachment description: mozharness_disable_l10n_vcs_sync.diff → Pause Gecko l10n vcs-sync
Attachment #8474668 - Attachment description: mozharness_enable_l10n_vcs_sync.diff → add new Gaia and l10n to vcs-sync
Attached patch aurora-in-tree-configs.diff (obsolete) — Splinter Review
Also bump B2G_UPDATE_CHANNEL
Attachment #8478516 - Attachment is obsolete: true
Attachment #8478516 - Flags: review?(hwine)
Attachment #8478529 - Flags: review?(hwine)
Attachment #8478532 - Flags: review?(hwine)
flame.xml and flame-kk.xml refer kernel_lm  using "firefox-one" and firefox-one-v1.4" branches what confuses/brakes branching. Preferring "firefox-one" since we don't use flame-kk in our configs.
Attachment #8478716 - Flags: review?(hwine)
(In reply to Rail Aliiev [:rail] from comment #23)
> flame.xml and flame-kk.xml refer kernel_lm  using "firefox-one" and
> firefox-one-v1.4" branches what confuses/brakes branching. Preferring
> "firefox-one" since we don't use flame-kk in our configs.

Fairly soon we will have flame-kk, bug 1055305.
Comment on attachment 8478431 [details] [diff] [review]
merge-mozharness-1.diff

Review of attachment 8478431 [details] [diff] [review]:
-----------------------------------------------------------------

lgtm
Attachment #8478431 - Flags: review?(hwine) → review+
Attachment #8478434 - Flags: review?(hwine) → review+
Comment on attachment 8478529 [details] [diff] [review]
aurora-in-tree-configs.diff

Review of attachment 8478529 [details] [diff] [review]:
-----------------------------------------------------------------

all present and accounted for
Attachment #8478529 - Flags: review?(hwine) → review+
Attachment #8478532 - Flags: review?(hwine) → review+
Comment on attachment 8478716 [details] [diff] [review]
b2g_branch_repos.py.diff

Review of attachment 8478716 [details] [diff] [review]:
-----------------------------------------------------------------

good for this merge day

::: configs/merge_day/b2g_branch_repos.py
@@ +29,5 @@
>          "b2g-4.3_r2.1",   # prefer jellybean over kitkat for now, since
>          "b2g-jellybean",  # most of our builds use jb
>          "ics_chocolate_rb4.2",  # prefer wasabi's hardware_qcom_display over
>                                  # otoro's
> +        "foxfone-one",  # prefer flame's kernel_lm ove flame-kk's

do we need to open a bug to handle flame-kk per comment 25
Attachment #8478716 - Flags: review?(hwine) → review+
Attachment #8478774 - Flags: review?(hwine) → review+
Comment on attachment 8479470 [details] [diff] [review]
Add new Gaia

Review of attachment 8479470 [details] [diff] [review]:
-----------------------------------------------------------------

yay! doc update must have helped!
Attachment #8479470 - Flags: review?(hwine) → review+
(In reply to Hal Wine [:hwine] (use needinfo) from comment #30)
> yay! doc update must have helped!

Yeah, thanks for the update!
Summary: Tracking bug for 01-sep-2014 migration work → Tracking bug for 02-sep-2014 migration work
Blocks: 1059451
(In reply to Hal Wine [:hwine] (use needinfo) from comment #29)
> do we need to open a bug to handle flame-kk per comment 25

Let me ask here first.

Viral,

in https://github.com/mozilla-b2g/b2g-manifest/commit/333d215958d5d12bb0a69f8a2c2a830462786e7c#diff-ca3d85637592a0cbef22a5d3e895f1d1R14 flame-kk started tracking foxfone-one-v1.4, while flame tracks the same repo using the foxfone-one branch.

This means, when we branch v2.1 off of master we have to use either foxfone-one-v1.4 or foxfone-one as a point on which v2.1 will be based on.

In other words if I branched today, v2.1 would look like this:
  $ git show-ref v2.1  origin/foxfone-one-v1.4 origin/foxfone-one
  9eb619d2efdf4bd121587d8296f5c10481f750b8 refs/heads/v2.1
  9eb619d2efdf4bd121587d8296f5c10481f750b8 refs/remotes/origin/foxfone-one
  9e62af4da848d56841bdde326f9bba26c743c33a refs/remotes/origin/foxfone-one-v1.4

Additionally, the new manifests would look like this:

flame-kk.xml (no changes):
...
<project name="kernel_lk" path="bootable/bootloader/lk" remote="b2g" revision="foxfone-one-v1.4"/>
...

flame.xml (changed from foxfone-one to v2.1):
...
<project name="kernel_lk" path="bootable/bootloader/lk" remote="b2g" revision="v2.1"/>
...

Any concerns regarding the changes above?
Flags: needinfo?(vwang)
(In reply to Rail Aliiev [:rail] from comment #32)
> (In reply to Hal Wine [:hwine] (use needinfo) from comment #29)
> > do we need to open a bug to handle flame-kk per comment 25
> 
> Let me ask here first.
> 
> Viral,
> 
> in
> https://github.com/mozilla-b2g/b2g-manifest/commit/
> 333d215958d5d12bb0a69f8a2c2a830462786e7c#diff-
> ca3d85637592a0cbef22a5d3e895f1d1R14 flame-kk started tracking
> foxfone-one-v1.4, while flame tracks the same repo using the foxfone-one
> branch.
> 
> This means, when we branch v2.1 off of master we have to use either
> foxfone-one-v1.4 or foxfone-one as a point on which v2.1 will be based on.
> 
> In other words if I branched today, v2.1 would look like this:
>   $ git show-ref v2.1  origin/foxfone-one-v1.4 origin/foxfone-one
>   9eb619d2efdf4bd121587d8296f5c10481f750b8 refs/heads/v2.1
>   9eb619d2efdf4bd121587d8296f5c10481f750b8 refs/remotes/origin/foxfone-one
>   9e62af4da848d56841bdde326f9bba26c743c33a
> refs/remotes/origin/foxfone-one-v1.4
> 
> Additionally, the new manifests would look like this:
> 
> flame-kk.xml (no changes):
> ...
> <project name="kernel_lk" path="bootable/bootloader/lk" remote="b2g"
> revision="foxfone-one-v1.4"/>
> ...
> 
> flame.xml (changed from foxfone-one to v2.1):
> ...
> <project name="kernel_lk" path="bootable/bootloader/lk" remote="b2g"
> revision="v2.1"/>
> ...
> 
> Any concerns regarding the changes above?
I think we should use the branch name like this:
refs/remotes/origin/foxfone-one => flame-jb
refs/remotes/origin/foxfone-one-v1.4 => flame-kk

Since partner only release jb+v1.3 and kk+v1.4 to us, they use foxfone-one-v1.4 for kk build.

I think reference phone need to support different combinations like:
jb+v1.4, jb+v2.0, jb+v2.1, kk+v1.4, kk+v2.0, kk+v2.1
for those jb base images, we will use refs/remotes/origin/foxfone-one
for those kk base images, we will use refs/remotes/origin/foxfone-one-v1.4

Hi Michael,

Any feedback in your side?
Thank you!
Flags: needinfo?(vwang) → needinfo?(mwu)
Does this mean that we shouldn't even create v2.1 branch in kernel_lk and continue using existing branches for kernel_lk in flame.xml and flame-kk.xml?
I don't think we need to care about kk+v1.4. Also, jb+v2.1 can probably be ignored if the base images for KK are publicly released soon.
Flags: needinfo?(mwu)
Attachment #8481307 - Flags: review?(hwine)
Attachment #8474668 - Attachment is obsolete: true
Attachment #8481313 - Flags: review?(hwine)
Attached patch Branch gaia too (obsolete) — Splinter Review
* Gaia needs to be branched, removing it from no_branch_repos and extra_branch_manifest_repos
* no_lock_revision_repos is not used anywhere, removing
Attachment #8481335 - Flags: review?(hwine)
Comment on attachment 8481313 [details] [diff] [review]
[new vcssync] start syncing geckco l10n (m-a -> v2.1, m-c ->v2.2)

Review of attachment 8481313 [details] [diff] [review]:
-----------------------------------------------------------------

lgtm for gecko
Attachment #8481313 - Flags: review?(hwine) → review+
Comment on attachment 8481307 [details] [diff] [review]
[new vcssync] start syncing gaia l10n from central to v2.1

Review of attachment 8481307 [details] [diff] [review]:
-----------------------------------------------------------------

r- -- this isn't for now, but when l10n cuts over
Attachment #8481307 - Flags: review?(hwine) → review-
Comment on attachment 8481335 [details] [diff] [review]
Branch gaia too

Review of attachment 8481335 [details] [diff] [review]:
-----------------------------------------------------------------

<stamp/>
Attachment #8481335 - Flags: review?(hwine) → review+
It turns out I need to leave "gaia" in extra_branch_manifest_repos to avoid locking it to a particular revision in b2g-manifests. Works "on my laptop". :)
Attachment #8481335 - Attachment is obsolete: true
Attachment #8481515 - Flags: review?(hwine)
Comment on attachment 8481307 [details] [diff] [review]
[new vcssync] start syncing gaia l10n from central to v2.1

Obsoleting this
Attachment #8481307 - Attachment is obsolete: true
Comment on attachment 8481515 [details] [diff] [review]
branch_gaia_too.diff

Review of attachment 8481515 [details] [diff] [review]:
-----------------------------------------------------------------

<stamp/>
Attachment #8481515 - Flags: review?(hwine) → review+
something landed into production
Adds flame-kk to your patch.
Attachment #8478529 - Attachment is obsolete: true
Attachment #8482085 - Flags: review?(rail)
Adds recent additions (dolphin and flame-kk).
Attachment #8478431 - Attachment is obsolete: true
Attachment #8482087 - Flags: review?(rail)
Comment on attachment 8478286 [details] [diff] [review]
merge-day-buildbot-configs.diff

This will conflict with bug 1055305, fun times. We want to make sure we don't enable nightlies for flame-kk yet.
Attachment #8482085 - Flags: review?(rail) → review+
Attachment #8482087 - Flags: review?(rail) → review+
Attachment #8482674 - Flags: review?(standard8)
Comment on attachment 8482674 [details] [diff] [review]
tb_merge_day.diff

Great, thanks!
Attachment #8482674 - Flags: review?(standard8) → review+
Comment on attachment 8481313 [details] [diff] [review]
[new vcssync] start syncing geckco l10n (m-a -> v2.1, m-c ->v2.2)

remote:   https://hg.mozilla.org/build/mozharness/rev/5675ff8803dd
remote:   https://hg.mozilla.org/build/mozharness/rev/a7ce19755ef4
Attachment #8481313 - Flags: checked-in+
Apparently it takes a regular Bumper Bot push to point Gaia at the correct repo post-uplift:
https://hg.mozilla.org/releases/mozilla-aurora/rev/9698450944cd

This may explain the rash of random bustage we had in the pushes prior to that. Is there any reason we couldn't manually make the needed gaia.json changes along with theother post-uplift changes already being made like in the push below?
https://hg.mozilla.org/releases/mozilla-aurora/rev/5b8210dcf52a
I doubt that it would help if we changed the URL manually and left the changeset the same (because we branched at that point). Or there is something that I'm missing?
Update the changeset too?
(In reply to Ryan VanderMeulen [:RyanVM UTC-4] from comment #65)
> Apparently it takes a regular Bumper Bot push to point Gaia at the correct
> repo post-uplift:
> https://hg.mozilla.org/releases/mozilla-aurora/rev/9698450944cd
> 
> This may explain the rash of random bustage we had in the pushes prior to
> that. Is there any reason we couldn't manually make the needed gaia.json
> changes along with theother post-uplift changes already being made like in
> the push below?
> https://hg.mozilla.org/releases/mozilla-aurora/rev/5b8210dcf52a

One of the theories I have:

1) branching of the b2g repos happens
2) gaia v2.1 starts the conversion to intergration/gaia-2_1
3) central to aurora merge happens, gaia.json uses 14400ee07e836d74039e2317feebc0a13fcb60c8
4) intergration/gaia-2_1 is finally ready (it took multiple hours after step 2)
5) the bumper detect the bump way behind (there are ~70 changesets between 14400ee07e836d74039e2317feebc0a13fcb60c8 and 1e5e149934e3eb86064575e4bdf0bdef8d39e67b)

Hal may have some ideas too.
Flags: needinfo?(hwine)
Merged to production, and deployed. (yesterday)
I'm not familiar with bumper requirements and assumptions in detail. I do suspect that the several hour delay required to create the integration/gaia-2_1 hg repo would throw it off.

Rail & I have discussed ways to shorten the conversion time next time. That should help.

I do note that the phrase "random bustage we had in the pushes prior to [uplift]" bothers me a bit. I don't know if the design of the bumper is that "mid uplift" builds are valid. The conservative approach would be to ensure they don't occur.
Flags: needinfo?(hwine)
Status: NEW → RESOLVED
Closed: 5 years ago
Resolution: --- → FIXED
Status: RESOLVED → REOPENED
Resolution: FIXED → ---
Attachment #8493711 - Flags: review?(rail) → review+
Comment on attachment 8493711 [details] [diff] [review]
[new vcssync]: updating l10n staging config

On default branch:     https://hg.mozilla.org/build/mozharness/rev/f12caf1970aa
On production branch:  https://hg.mozilla.org/build/mozharness/rev/83effe9ef7fc
Attachment #8493711 - Flags: checked-in+
Status: REOPENED → RESOLVED
Closed: 5 years ago5 years ago
Resolution: --- → FIXED
Component: General Automation → General
You need to log in before you can comment on or make changes to this bug.