1474254 - GTest is permafailing on Windows ccov builds because it runs out of time

Status: RESOLVED FIXED

(Testing :: Code Coverage, defect)

Description

[Marco Castelluccio [:marco]]

GTest is always running over 1200 seconds, which is the max run time for GTest.

I see "TEST-START Jemalloc.JunkPoison" then several "INFO -  Running GTest tests..." followed by several "INFO -  Finished running GTest tests.".

Comment 1

[Marco Castelluccio [:marco]]

Do you know what could have changed recently to make this test take this much time?

Comment 2

[Marco Castelluccio [:marco]]

(In reply to Marco Castelluccio [:marco] from comment #1)
> Do you know what could have changed recently to make this test take this
> much time?

Actually setting needinfo now...

Comment 3

[Marco Castelluccio [:marco]]

Attachment: log.tar.xz

Comment 4

[Mike Hommey [:glandium]]

I assume you mean "code coverage windows builds", because they're pretty much green on all tier-1 windows builds. The test already takes a long time on the tier-1 windows builds. I guess code coverage builds are built with no optimization, and that would likely make them so much slower that they hit the time out. I don't think there's much that can be done here besides disabling the test or increasing the timeout.

Comment 5

[Marco Castelluccio [:marco]]

(In reply to Mike Hommey [:glandium] from comment #4)
> I assume you mean "code coverage windows builds", because they're pretty
> much green on all tier-1 windows builds. The test already takes a long time
> on the tier-1 windows builds. I guess code coverage builds are built with no
> optimization, and that would likely make them so much slower that they hit
> the time out. I don't think there's much that can be done here besides
> disabling the test or increasing the timeout.

Yep, I meant coverage builds. They have optimizations enabled, but are slower than normal debug builds anyways.
It's strange though that these tests didn't timeout for months and started consistently timing out only recently.
Would you prefer disabling the test or increasing the timeout?

Comment 6

[Mike Hommey [:glandium]]

If disabling is possible without too much hassle, disable.

Comment 7

[Calixte Denizet (:calixte)]

Attachment: Bug 1474254 -- Disable test Jemalloc.JunkPoison for Windows ccov builds.

GTest is permafailing on Windows because of timeout.

Comment 8

[Phabricator Automation]

Comment on attachment 8990940 [details]
Bug 1474254 -- Disable test Jemalloc.JunkPoison for Windows ccov builds.

Mike Hommey [:glandium] has approved the revision.

https://phabricator.services.mozilla.com/D2043

Comment 9

[Pulsebot]

Pushed by [e-mail]:
https://hg.mozilla.org/integration/mozilla-inbound/rev/a52f20385008
Disable test Jemalloc.JunkPoison for Windows ccov builds.

Comment 10

[Raul Gurzau (:RaulG)]

https://hg.mozilla.org/mozilla-central/rev/a52f20385008