Closed
Bug 1319824
Opened 9 years ago
Closed 8 years ago
Enable funsize for date
Categories
(Release Engineering :: General, defect)
Release Engineering
General
Tracking
(Not tracked)
RESOLVED
FIXED
People
(Reporter: rail, Assigned: sfraser)
Details
Attachments
(3 files, 2 obsolete files)
Builds on date are TC based, so we'll need to modify funsize to treat them differently. We should track the signed binaries only.
Assignee | ||
Comment 1•9 years ago
|
||
How do we track which jobs (or pulse messages) are related to the production of a signed binary?
Assignee | ||
Comment 2•9 years ago
|
||
In order to easily track the signing jobs, adding a custom pulse message route to the task graph seems appropriate. Funsize doesn't have to have any special logic to work out if it it's a task that needs listening to - the signing tasks will be the only ones publishing on this route.
Based on https://docs.taskcluster.net/manual/devel/namespaces it seems like a route name of `project.releng.funsize.level-3` is appropriate. Would that be an acceptable route name for these pulse messages?
Flags: needinfo?(rail)
Flags: needinfo?(bugspam.Callek)
Reporter | ||
Comment 3•9 years ago
|
||
project.releng.funsize.level-3 sounds good to me. I'm not sure what our naming practise is, but maybe you need to add some other info into the route, like the branch?
Flags: needinfo?(rail)
Assignee | ||
Comment 4•9 years ago
|
||
https://docs.taskcluster.net/manual/devel/namespaces has the naming guidelines. Adding in the branch sounds like a good idea. I'll put in a review request for 'project.releng.funsize.date.level-3'
Comment hidden (mozreview-request) |
Comment 6•9 years ago
|
||
(In reply to Simon Fraser [:sfraser] ⌚️GMT from comment #2)
> In order to easily track the signing jobs, adding a custom pulse message
> route to the task graph seems appropriate. Funsize doesn't have to have any
> special logic to work out if it it's a task that needs listening to - the
> signing tasks will be the only ones publishing on this route.
>
> Based on https://docs.taskcluster.net/manual/devel/namespaces it seems like
> a route name of `project.releng.funsize.level-3` is appropriate. Would that
> be an acceptable route name for these pulse messages?
I think we should probably meet about this, I'm not sure why we need a route/namespace, or why it makes sense for this case.
That said, I am planning on coercing the index routes to be on the signing task and not the unsigned task at some point in the (near) future though.
Flags: needinfo?(bugspam.Callek)
Assignee | ||
Comment 7•9 years ago
|
||
If there's a different, reliable way of identifying just the signing tasks from the pulse messages, we could use that. I'm trying to avoid having the funsize scheduler perform lots of API calls in order to identify which messages it needs to act on.
Comment hidden (mozreview-request) |
Assignee | ||
Updated•9 years ago
|
Attachment #8820676 -
Attachment is obsolete: true
Assignee | ||
Updated•9 years ago
|
Attachment #8820755 -
Attachment is obsolete: true
Comment 9•9 years ago
|
||
(In reply to Justin Wood (:Callek) from comment #6)
> (In reply to Simon Fraser [:sfraser] ⌚️GMT from comment #2)
> > In order to easily track the signing jobs, adding a custom pulse message
> > route to the task graph seems appropriate. Funsize doesn't have to have any
> > special logic to work out if it it's a task that needs listening to - the
> > signing tasks will be the only ones publishing on this route.
> >
> > Based on https://docs.taskcluster.net/manual/devel/namespaces it seems like
> > a route name of `project.releng.funsize.level-3` is appropriate. Would that
> > be an acceptable route name for these pulse messages?
>
> I think we should probably meet about this, I'm not sure why we need a
> route/namespace, or why it makes sense for this case.
We chatted in IRC about this today, takeaways:
* partials are tier1 blocking
* this is one of the easier ways we envision being able to identify which tasks funsize needs to act on.
Because of that, I'm not going to block on a thought-project on if there is a better way to go about this, a route isn't that bad of a thing.
Comment hidden (mozreview-request) |
Comment 11•9 years ago
|
||
mozreview-review |
Comment on attachment 8820793 [details]
bug 1319824 Announce signing jobs to funsize with
https://reviewboard.mozilla.org/r/100232/#review100722
Also, please put the commit message on one line, otherwise `hg log` and summary of such won't show the whole thing, leaving a shortened actual message in consoles.
::: taskcluster/taskgraph/transforms/signing.py:67
(Diff revision 1)
> # taskcluster/taskgraph/transforms/task.py for the schema details, and the
> # below transforms for defaults of various values.
> Optional('treeherder'): task_description_schema['treeherder'],
> +
> + # If True, announce to a specific route when done, so that funsize
> + # can start its partial mar generation. For nightly builds only.
very nitpicky:
"If True, adds a route which funsize uses to schedule generation of partial mar files for updates. Expected to be added on nightly builds only"
::: taskcluster/taskgraph/transforms/signing.py:124
(Diff revision 1)
> 'attributes': attributes,
> 'run-on-projects': dep_job.attributes.get('run_on_projects'),
> 'treeherder': treeherder,
> }
>
> + if job.get('use_funsize_route', False):
can omit the ", False" but doesn't hurt to have it.
::: taskcluster/taskgraph/transforms/signing.py:125
(Diff revision 1)
> 'run-on-projects': dep_job.attributes.get('run_on_projects'),
> 'treeherder': treeherder,
> }
>
> + if job.get('use_funsize_route', False):
> + task['routes'] = ["project.releng.funsize.{project}.level-3".format(project=config.params['project'])]
nit: also format "level" so:
"project.releng.funsize.{project}.level-{level}".format(... level=config.params['level'])
Comment 12•9 years ago
|
||
mozreview-review |
Comment on attachment 8820793 [details]
bug 1319824 Announce signing jobs to funsize with
https://reviewboard.mozilla.org/r/100232/#review100724
Attachment #8820793 -
Flags: review?(bugspam.Callek) → review+
Comment hidden (mozreview-request) |
Comment 14•9 years ago
|
||
sfraser:
in the future you can do `hg commit --amend` to modify an existing commit with new content and/or change the commit message.
You can also use `hg histedit` to do things like fold two commits together and select certain commits to modify a commit message for as well, incase that workflow works better for you.
In this case, I landed to date, but I folded the two commits together, and fixed the commit message while I was at it...
https://hg.mozilla.org/projects/date/rev/a331041b33dab6eceaafd29e4666042d9a5def37
Status: NEW → RESOLVED
Closed: 9 years ago
Resolution: --- → FIXED
Assignee | ||
Comment 15•9 years ago
|
||
mozreview-review |
Comment on attachment 8820815 [details]
based on review, change comment and add config-based level in the route
https://reviewboard.mozilla.org/r/100234/#review102238
All done
Assignee | ||
Comment 16•9 years ago
|
||
https://github.com/mozilla-releng/funsize/pull/42 up for the main piece of work.
I didn't notice this had been marked as resolved/fixed. Reopened until the work is done
Status: RESOLVED → REOPENED
Resolution: FIXED → ---
Assignee | ||
Updated•9 years ago
|
Assignee: nobody → sfraser
Comment hidden (mozreview-request) |
Comment 18•9 years ago
|
||
mozreview-review |
Comment on attachment 8825060 [details]
bug 1319824 Bump funsize version for pulse route name reordering
https://reviewboard.mozilla.org/r/103312/#review104130
Didn't follow-up carefully the pulse BB/TC conversation but from puppet standpoint looks good to me.
Comment 19•9 years ago
|
||
Comment on attachment 8825060 [details]
bug 1319824 Bump funsize version for pulse route name reordering
https://hg.mozilla.org/build/puppet/rev/0b685510abdf
Attachment #8825060 -
Flags: checked-in+
Comment hidden (mozreview-request) |
Comment 21•9 years ago
|
||
mozreview-review |
Comment on attachment 8825060 [details]
bug 1319824 Bump funsize version for pulse route name reordering
https://reviewboard.mozilla.org/r/103312/#review105320
Updated•9 years ago
|
Attachment #8825060 -
Flags: review?(aki)
Updated•9 years ago
|
Attachment #8825060 -
Flags: review?(aki) → review+
Comment 22•8 years ago
|
||
Anything left to do here?
Assignee | ||
Comment 23•8 years ago
|
||
With the funsize breakage the other day this is the first day I've been able to verify that the partials actually work, and they do. As far as I can see this is all done
Reporter | ||
Comment 24•8 years ago
|
||
\o/ \o/ \o/
Thank you Simon!
Status: REOPENED → RESOLVED
Closed: 9 years ago → 8 years ago
Resolution: --- → FIXED
Updated•7 years ago
|
Component: General Automation → General
You need to log in
before you can comment on or make changes to this bug.
Description
•