1307673 - Request approval of stage credentials

Status: RESOLVED FIXED

(Tree Management :: Treeherder: API, defect)

Description

[Askeing Yen[:askeing][:fyen]]

We currently plan to push perf data to Perfherder.
And the client-id is 'hasal-dev'.

Thank you.

Comment 1

[Ed Morley [:emorley]]

Submitting jobs via the API is something we're trying to steer new projects away from in favour of submitting via Pulse queues instead.

There have also been a few projects that have started off trying to schedule their own jobs/upload logs to S3/submit to treeherder/..., only to find that it's quite a bit of hassle, and so later switch to using Taskcluster to do the scheduling etc for them.

Have you looked into using Taskcluster? If it's viable for what you're after, it will cut down the amount of work required considerably :-)

Comment 2

[William Lachance (:wlach)]

(In reply to Ed Morley [:emorley] from comment #1)
> Submitting jobs via the API is something we're trying to steer new projects
> away from in favour of submitting via Pulse queues instead.
> 
> There have also been a few projects that have started off trying to schedule
> their own jobs/upload logs to S3/submit to treeherder/..., only to find that
> it's quite a bit of hassle, and so later switch to using Taskcluster to do
> the scheduling etc for them.
> 
> Have you looked into using Taskcluster? If it's viable for what you're
> after, it will cut down the amount of work required considerably :-)

I suspect Hasal doesn't/can't use a docker worker (a performance tool usually needs to run on bare metal), so I don't think Taskcluster is currently an option for them. I think submitting data via the python client is probably the best and easiest method, in this case. Hopefully using something like autophone (https://github.com/mozilla/autophone/blob/master/autophonetreeherder.py) as an example, it won't be so bad.

Comment 3

[William Lachance (:wlach)]

I went ahead and approved these credentials. It's only stage, can't hurt that much. :)