Closed
Bug 1406517
Opened 7 years ago
Closed 7 years ago
fennec release bouncer sub in-tree task
Categories
(Release Engineering :: Release Automation: Other, enhancement)
Release Engineering
Release Automation: Other
Tracking
(Not tracked)
RESOLVED
FIXED
People
(Reporter: mozilla, Assigned: mozilla)
References
Details
Attachments
(5 files, 2 obsolete files)
Let's build a bbb task in-tree for bouncer submission.
Ideally this will set up the framework for other bbb tasks to follow, and will allow for firefox bouncer submission as well.
Assignee | ||
Comment 1•7 years ago
|
||
Assignee | ||
Comment 2•7 years ago
|
||
Assignee | ||
Comment 3•7 years ago
|
||
Rail, I'm wondering what's problematic in here.
Do we rely on specific routes?
We probably need to add payload.properties release_promotion, repo_path, and script_repo_revision.
We also don't have anything other than 'parent' in task.extra. Are task.extra.build_props used? I imagine we'll either want to redo notifications, or add task.extra.notifications back in in another transform.
Attachment #8916118 -
Flags: feedback?(rail)
Assignee | ||
Comment 4•7 years ago
|
||
Callek, how does this in-tree patch look?
It'll be applied against maple, which has diverged a bit from m-c.
Attachment #8916120 -
Flags: feedback?(bugspam.Callek)
Comment 5•7 years ago
|
||
Comment on attachment 8916120 [details] [diff] [review]
bouncer-sub.patch
Review of attachment 8916120 [details] [diff] [review]:
-----------------------------------------------------------------
::: taskcluster/taskgraph/transforms/job/buildbot.py
@@ +62,2 @@
> })
> + worker.setdefault('properties', {}).update(props)
personally I would love if we had a different transform to inject release stuff here, *but* given the knowledge the BBB is on its last legs, I'm ok with this rather than an over-engineered solution
Attachment #8916120 -
Flags: feedback?(bugspam.Callek) → feedback+
Comment 6•7 years ago
|
||
Comment on attachment 8916118 [details]
dictdiffer diff of 57.0b3 and in-tree tasks
(In reply to Aki Sasaki [:aki] from comment #3)
> Created attachment 8916118 [details]
> dictdiffer diff of 57.0b3 and in-tree tasks
>
> Rail, I'm wondering what's problematic in here.
* Removing the scopes probably will brake BBB: https://github.com/mozilla-releng/buildbot-bridge/blob/master/bbb/services.py#L488
* payload.properties.build_number should be an integer. IIRC we use it in some calculations.
BTW, the schema is at https://github.com/mozilla-releng/buildbot-bridge/blob/master/bbb/schemas/payload.yml
> Do we rely on specific routes?
Probably not for this particular builder, but some of the QE tests listen for these indexes (at least to trigger desktop update tests).
> We probably need to add payload.properties release_promotion, repo_path, and
> script_repo_revision.
I think so too. They are used to dynamically pass the info to the builder.
>
> We also don't have anything other than 'parent' in task.extra. Are
> task.extra.build_props used? I imagine we'll either want to redo
> notifications, or add task.extra.notifications back in in another transform.
I think this is OK. Let's not cargo cult things unless we need them. :)
Attachment #8916118 -
Flags: feedback?(rail) → feedback+
Assignee | ||
Comment 7•7 years ago
|
||
(In reply to Rail Aliiev [:rail] ⌚️ET from comment #6)
> * payload.properties.build_number should be an integer. IIRC we use it in
> some calculations.
Ah, I changed it to a string because I was hitting [1], which says all properties in the payload need to be a taskref_or_string. Then I tried adding `Optional("build_number"): int,` and hit "YAML schemas should use dashed lower-case identifiers" [2].
Rail: I'm guessing sending a string `"build_number": "1"` could be translated to an int() at some point downstream in bbb or buildbot -- is that a pain? Is that more or less of a pain than sending a `"build-number": 1` that we'll need to translate into a `build_number` ?
Dustin: is there a less-ugly way we could relax the schema for buildbot-bridge tasks only, with the understanding that buildbot-bridge usage is scheduled to be retired by Fx59? I suppose I could add `taskref_or_string_or_int` for `payload.Extra`...
[1] https://hg.mozilla.org/mozilla-central/file/a0488ecc201c/taskcluster/taskgraph/transforms/task.py#l372
[2] https://hg.mozilla.org/mozilla-central/file/tip/taskcluster/taskgraph/util/schema.py#l165
Flags: needinfo?(rail)
Flags: needinfo?(dustin)
Assignee | ||
Comment 8•7 years ago
|
||
Ah, we might be able to use WHITELISTED_SCHEMA_IDENTIFIERS.
Flags: needinfo?(dustin)
Assignee | ||
Comment 10•7 years ago
|
||
This one looks good...
Attachment #8916116 -
Attachment is obsolete: true
Assignee | ||
Comment 11•7 years ago
|
||
Attachment #8916118 -
Attachment is obsolete: true
Assignee | ||
Comment 12•7 years ago
|
||
We're missing "project:releng:buildbot-bridge:builder-name:release-maple-fennec_bncr_sub", and
"queue:route:index.releases.v1.maple.latest.fennec.latest.bouncer_submitter".
I'm going to guess we want
project:releng:buildbot-bridge:*
queue:route:index.releases.v1.maple.* on maple.
Assignee | ||
Comment 13•7 years ago
|
||
(In reply to Aki Sasaki [:aki] from comment #12)
> We're missing
> "project:releng:buildbot-bridge:builder-name:release-maple-fennec_bncr_sub",
> and
> "queue:route:index.releases.v1.maple.latest.fennec.latest.bouncer_submitter".
>
> I'm going to guess we want
>
> project:releng:buildbot-bridge:*
> queue:route:index.releases.v1.maple.* on maple.
If it helps, we could add an assume role that releng has access to modify, and add that role to maple, so we could make further maple changes in the future.
Assignee | ||
Comment 14•7 years ago
|
||
Attachment #8917574 -
Flags: review?(rail)
Updated•7 years ago
|
Attachment #8917574 -
Flags: review?(rail) → review+
Assignee | ||
Comment 15•7 years ago
|
||
Fixed on maple; bug 1412690 tracks landing on m-c.
Status: NEW → RESOLVED
Closed: 7 years ago
Resolution: --- → FIXED
You need to log in
before you can comment on or make changes to this bug.
Description
•