Closed Bug 997946 Opened 10 years ago Closed 10 years ago

Stop running cppunit and jittest on OS X 10.8 on all branches

Categories

(Release Engineering :: General, defect)

x86_64
macOS
defect
Not set
normal

Tracking

(Not tracked)

RESOLVED FIXED

People

(Reporter: philor, Assigned: dminor)

References

Details

(Whiteboard: [capacity])

Attachments

(3 files, 1 obsolete file)

It was a nice little bonus that moving the cpp and jittests off the buildslaves also got them running on the multiple test platforms like WinXP/Win7/Win8 instead of just Server 2008, but in the case of 10.8 it's a bonus we can't afford. We're at a 20 hour backlog right now, up from a typical peak of 10-12 hours before we added these and made browser-chrome so slow we had to split it into four chunks and endure the extra overhead of running it in more chunks.

Nobody actually asked for coverage on 10.8, so it's my contention that you don't have to wait weeks while trying to persuade someone that they are the stakeholder who should decide it's okay to shut them off. It's an emergency, just kill them, 10.6 is plenty of OS X coverage for them.
Whiteboard: [capacity]
I think this will do what we want.
Assignee: nobody → dminor
Status: NEW → ASSIGNED
Attachment #8422474 - Flags: review?(kmoir)
Attached file bug997946.diff
builder diff
Comment on attachment 8422474 [details] [diff] [review]
Disable cppunit and jittest for mountainlion

dminor it looks good although I notice that these tests are still enabled on esr and some other m-r etc branches because it's in a loop to ride the trains.  I don't think this matters much though because these branches don't generate many jobs compared the the main ones.
Attachment #8422474 - Flags: review?(kmoir) → review+
This patch actually seems like it made 10.8 CppUnit tests run on ESR 24, which it promptly fails to do.

According to https://tbpl.mozilla.org/?tree=Mozilla-Esr24&showall=1&jobname=cppunit (if you click the green down arrow a few times), Cpp has never run separately there... So now it's just spinning up 10.8 machines just to let them die a few minutes later.
Flags: needinfo?(dminor)
Yes, this is my fault, the logic for handling the cppunit tests is actually the reverse of the jittests - it removes tests from suites instead of adding them (consistency is overrated). I'll fix it right away.
Flags: needinfo?(dminor)
Attachment #8426214 - Flags: review?(bhearsum)
Attachment #8426214 - Flags: review?(bhearsum) → review+
(In reply to Dan Minor [:dminor] from comment #8)
> Thanks pushed to:
> https://hg.mozilla.org/build/buildbot-configs/rev/b06d5a4eb728

That failed the unittests, backed out here: https://hg.mozilla.org/build/buildbot-configs/rev/c5bd391061f9
Take 2. I don't have access to the vpn to see the unittest failures, so I'm not certain what failed the first time. I've made the loop closer to the original loop to hopefully avoid the problem(s). If it happens again, I'll track down what is failing.

I also noticed that we were running a loop over suite_type but then not actually using the loop variable when removing the suites, so I've fixed that as well.
Attachment #8426214 - Attachment is obsolete: true
Attachment #8427085 - Flags: review?(bhearsum)
Attachment #8427085 - Flags: review?(bhearsum) → review+
Something here went live today
Merged and deployed to production.
Blocks: 1013683
Status: ASSIGNED → RESOLVED
Closed: 10 years ago
Resolution: --- → FIXED
Component: General Automation → General
You need to log in before you can comment on or make changes to this bug.

Attachment

General

Created:
Updated:
Size: