1137694 - Fix and re-enable browser_net_sort-03.js on Linux opt e10s

Status: RESOLVED FIXED

(DevTools :: Netmonitor, defect)

Description

[Ryan VanderMeulen [:RyanVM]]

For some reason, browser_net_sort-03.js started failing ~50% of the time yesterday. Numerous attempts at backouts and retriggers have yielded no progress. Due to this causing an ongoing tree closure, we're disabling the test to get the trees reopened. This bug tracks finding the root cause and getting it re-enabled.

Comment 1

[Ryan VanderMeulen [:RyanVM]]

Attachment: test log

Comment 2

[Ryan VanderMeulen [:RyanVM]]

Attachment: test screenshot

Comment 3

[Ryan VanderMeulen [:RyanVM]]

Disabled:
https://hg.mozilla.org/integration/mozilla-inbound/rev/c236c18ab5c8

Comment 4

[Patrick Brosset <:pbro>]

No knowing the netmonitor test well, I can't investigate the root cause, also it looks like it's more related to recent e10s changes rather than devtools changes, because there hasn't been anything that seems related lately.
What I *can* do however, is change the test so it doesn't directly manipulate the content window CPOW anymore.
The timeout occurs right when trying to call a content window method directly, and the logs say that the CPOW is dead. This is weird, but going via a message to the frame script instead should help.
I will upload a patch in a second.

Comment 5

[Ryan VanderMeulen [:RyanVM]]

CCing some e10s folks as well.

Comment 6

[Patrick Brosset <:pbro>]

Attachment: MozReview Request: bz://1137694/pbrosset

/r/4427 - Bug 1137694 - Avoid calling content methods directly in browser_net_sort-03.js; r=past

Pull down this commit:

hg pull review -r 89f3376842da3dc6391ba434874949ead53abdb8

Comment 7

[Panos Astithas (he/him) [:past] (please ni?)]

Comment on attachment 8570503 [details]
MozReview Request: bz://1137694/pbrosset

https://reviewboard.mozilla.org/r/4425/#review3603

::: browser/devtools/netmonitor/test/browser_net_sort-03.js
(Diff revision 1)
> -        .then(finish);
> -    });
> +        .then(finish, Cu.reportError);
> +    }, Cu.reportError);

Is calling Cu.reportError in browser mochitests a thing now? I'm more used to see ok(false, ...) in these cases, but if it works, that's fine.

Comment 8

[Panos Astithas (he/him) [:past] (please ni?)]

Comment on attachment 8570503 [details]
MozReview Request: bz://1137694/pbrosset

https://reviewboard.mozilla.org/r/4425/#review3605

Ship It!

Comment 9

[Ryan VanderMeulen [:RyanVM]]

https://hg.mozilla.org/mozilla-central/rev/c236c18ab5c8

Comment 10

[Patrick Brosset <:pbro>]

https://reviewboard.mozilla.org/r/4427/#review3673

Ship It!

Comment 11

[Patrick Brosset <:pbro>]

Corrected the patch as per Panos' comments, and re-enabled the test.
Carried over the R+.
Pushed to try: https://treeherder.mozilla.org/#/jobs?repo=try&revision=6c30df15f7d5

Comment 12

[Patrick Brosset <:pbro>]

Of course I didn't run the test locally before pushing to try and just realized that another test was using the same test HTML page I modified and now therefore fails.
I'll get this other test fixed now.
The good news is that so far, I'm not seeing browser_net_sort-03.js failing on linux opt e10s.

Comment 13

[Patrick Brosset <:pbro>]

https://reviewboard.mozilla.org/r/4427/#review3675

Ship It!

Comment 14

[Patrick Brosset <:pbro>]

https://treeherder.mozilla.org/#/jobs?repo=try&revision=5ded0a9ebf9c

Comment 15

[Ryan VanderMeulen [:RyanVM]]

(In reply to Patrick Brosset [:pbrosset] [:patrick] from comment #12)
> Of course I didn't run the test locally before pushing to try and just
> realized that another test was using the same test HTML page I modified and
> now therefore fails.
> I'll get this other test fixed now.
> The good news is that so far, I'm not seeing browser_net_sort-03.js failing
> on linux opt e10s.

I don't see an updated rev on MozReview?

Comment 16

[Patrick Brosset <:pbro>]

I might be missing something about MozReview ... I did push to review and can see the right diff on the MozReview website.
Anyway, pushed to fx-team: https://hg.mozilla.org/integration/fx-team/rev/37409aced15c

Comment 17

[Wes Kocher (:KWierso) (Not reading bugmail; email directly if needed)]

https://hg.mozilla.org/mozilla-central/rev/37409aced15c

Comment 19

[Patrick Brosset <:pbro>]

Attachment: MozReview Request: Bug 1137694 - Avoid calling content methods directly in browser_net_sort-03.js; r=past