1295840 - Unhide e10s web-platform-tests when they stop constantly timing out in utterly random tests

Status: RESOLVED FIXED

(Tree Management Graveyard :: Visibility Requests, defect)

Description

[Phil Ringnalda (:philor)]

I tried to star them all; I've filed 233 intermittent timeout w-p-t bugs this month, but now I'm tired, and I'm not anywhere close to reaching my goal, which was to get to where the average try push wouldn't wind up hitting 3-5 unfiled intermittent failures. Looks to me like just getting rid of debug e10s will be enough to get the problem down to a reasonable level.

Hidden on trunk trees (except Ash, where I presume Ryan wants to keep seeing them fail) and try, we can unhide them when they no longer hit more than one unfiled timeout across all platforms over ten pushes.

Comment 1

[Phil Ringnalda (:philor)]

Turns out I was wrong about just debug being enough to make a tolerable situation: hidden on opt as well.

Comment 2

[James Graham [:jgraham]]

David, will it be possible to get some attention on this before I'm back on Monday? Otherwise I guess I will have to work on this urgently on Monday morning; losing wpt coverage on e10s seems rather unacceptable. I think the first line of attack should be retriggering back to the middle of July to look for the point at which the rate of intermittents ticks up.

Comment 3

[David Burns :automatedtester]

spoke to jgraham on IRC so clearing n-i

Comment 4

[James Graham [:jgraham]]

12:44 <@jonco> jgraham: yes, I think bug https://bugzilla.mozilla.org/show_bug.cgi?id=1288579 should have fixed things
12:44 < firebot> Bug 1288579 — FIXED, [e-mail] — Hang in js::GCParallelTask::join()

Can we unhide these now, please :)

Comment 5

[Ryan VanderMeulen [:RyanVM]]

Has anybody looked at TH to verify that?

Comment 6

[James Graham [:jgraham]]

Yes, it looks much better now. Mostly unexpected-fails on actual intermittent tests.