978507 - for mozharness desktop builds, replace usage of client.mk with upcoming mach/python script

Status: RESOLVED FIXED

(Release Engineering :: Applications: MozharnessCore, defect)

Comment 1

[Jordan Lund (:jlund)]

Attachment: 140706_bug_978211_proxy_script_mach_automation-mozharness.patch

This patch preps mozharness for 'mach build'. It won't land until latest
build-config changes make it to m-c.

Overview:

- fights some bitrot
- adds 'mach build' and supporting logic + removes un-needed make target calls
- adds configs for windows and mac
- changes how we deal with return_code and automation. it is only handled in
  summarize.
- partial mar is not included here (we will be using ffledgling's service
  eventually)
- snippets also are not included (we will just be using balrog)

this patch will only affect existing linux builders in cedar. will land a buildbot patch to turn on mac and windows after https://bugzilla.mozilla.org/show_bug.cgi?id=978211#c70 is fixed up.


log example of what this new build will look like for a linux32 opt build: http://dev-master1.srv.releng.scl3.mozilla.com:8037/builders/Linux%20mozilla-central%20build/builds/84/steps/run_script/logs/stdio

Comment 2

[Aki Sasaki (not active)]

Comment on attachment 8451678 [details] [diff] [review]
140706_bug_978211_proxy_script_mach_automation-mozharness.patch

r+ with a caveat.

>- partial mar is not included here (we will be using ffledgling's service
>  eventually)

Who was part of this decision?  If thought was put into this, I'll roll with it.  Aiui, however, we moved away from generating partials on a single machine and went with on-slave partials, to avoid having partial generation going well into the next day after a nightly, resulting in most people downloading full updates.  I haven't been following, but unless that's been resolved, that sounds like we may want to keep on-slave partials for the most common scenario(s) and use funsize for the next most common scenarios.

>- snippets also are not included (we will just be using balrog)

sgtm.

I'm largely *stamp*ing the changes, but didn't catch anything else egregious.

Comment 3

[Jordan Lund (:jlund)]

Attachment: 140708_bug_978211_proxy_script_mach_automation-mozharness.patch

r=aki

accidentally added a change I had in my testing env.

Comment 4

[Jordan Lund (:jlund)]

Comment on attachment 8452809 [details] [diff] [review]
140708_bug_978211_proxy_script_mach_automation-mozharness.patch

Review of attachment 8452809 [details] [diff] [review]:
-----------------------------------------------------------------

https://hg.mozilla.org/build/mozharness/rev/f2b8bf507e9a

Comment 5

[Jordan Lund (:jlund)]

Comment on attachment 8451678 [details] [diff] [review]
140706_bug_978211_proxy_script_mach_automation-mozharness.patch

Review of attachment 8451678 [details] [diff] [review]:
-----------------------------------------------------------------

I am submitting this patch as cedar does not run nightly anyway.

However, until the future of funsize and partial mar's are clarified, I will follow up in this bug with re-adding partial mar gen. Rather than sticking it into build configs, I will most likely add it to the 'update' action before we submit to balrog.

checked in: http://hg.mozilla.org/build/mozharness/rev/28a22339cb48

Comment 6

[Pete Moore [:pmoore][:pete]]

In production

Comment 7

[Jordan Lund (:jlund)]

this was done a while ago. I will follow up with some bugs to track its rolling out across remaining branches