1886588 - Print WPTs often have a 5-10 second delay between "Testing [URL]" and "Rendering Page 1/N", which causes many tests to timeout

Status: NEW

(Testing :: web-platform-tests, defect)

Description

[Daniel Holbert [:dholbert]]

Bug 1886581 clued me into the fact that we've got some suspicious delays in our WPT logs for print tests, at least in the log linked there.

Log (for an opt build with no obvious reason to be slow):
https://treeherder.mozilla.org/logviewer?job_id=451534097&repo=mozilla-beta&lineNumber=2669

Here's a snippet of that log, showing just a few pairs of consecutive lines where there's this supsicious delay in the timestamps between them (which is responsible for essentially the entire test duration):

18:45:50     INFO - PID 7124 | 1710960350188	Marionette	INFO	Testing http://web-platform.test:8000/css/CSS2/pagination/block-page-break-inside-avoid-15-print.html == http://web-platform.test:8000/css/CSS2/pagination/block-page-break-inside-avoid-15-print-ref.html
18:45:57     INFO - PID 7124 | 1710960357253	Marionette	INFO	Rendering page 1/5
[note, 7-second delay there]

18:47:45     INFO - PID 7124 | 1710960465719	Marionette	INFO	Testing http://web-platform.test:8000/css/CSS2/pagination/table-page-break-inside-avoid-5-print.html == http://web-platform.test:8000/css/CSS2/pagination/table-page-break-inside-avoid-5-print-ref.html
18:47:51     INFO - PID 7124 | 1710960471470	Marionette	INFO	Rendering page 1/4
[note, 6-second delay there]

We should see about either fixing these delays; or if they're not-easily-avoidable, we should increase the timeout threshold for this category of tests (to be substantially larger than the current 10s threshold), so that the delays don't cause spurious reports of intermittent test-timeouts.

Comment 1

[BugBot [:suhaib / :marco/ :calixte]]

The severity field is not set for this bug.
:jgraham, could you have a look please?

For more information, please visit BugBot documentation.

Comment 2

[Daniel Holbert [:dholbert]]

I did a quick pass through all "intermittent *-print.html" WPT test-failures that have been filed in the past ~6 months (since December), and they're all unexpected timeouts, likely due to this issue. I added them as dependencies of this bug.

(Many of them have been closed as INCOMPLETE, probably because this issue isn't test-specific; it affects all print tests to an equal extent, I imagine, with some amount of mysterious delay that's only flagged as a test failure if it happens to overshoot 10s which it occasionally does.)