This is failing quite frequent, it looks like almost permfailing on this path "accessible/tests/browser/"
This is an example, all the retriggers, except 1, failed: retriggers

I've gathered a few bugs in the "see also" section for more info.

Example of log snippet:

[task 2022-08-09T13:42:44.614Z] 13:42:44     INFO - TEST-START | accessible/tests/browser/e10s/browser_caching_value.js
[task 2022-08-09T13:42:44.928Z] 13:42:44     INFO - GECKO(1152) | must wait for focus in content
[task 2022-08-09T13:43:29.643Z] 13:43:29     INFO - TEST-INFO | started process screenshot
[task 2022-08-09T13:43:29.701Z] 13:43:29     INFO - TEST-INFO | screenshot: exit 0
[task 2022-08-09T13:43:29.702Z] 13:43:29     INFO - Buffered messages logged at 13:42:44
[task 2022-08-09T13:43:29.702Z] 13:43:29     INFO - Entering test bound 
[task 2022-08-09T13:43:29.703Z] 13:43:29     INFO - TEST-PASS | accessible/tests/browser/e10s/browser_caching_value.js | Actually remote browser - 
[task 2022-08-09T13:43:29.703Z] 13:43:29     INFO - Buffered messages finished
[task 2022-08-09T13:43:29.704Z] 13:43:29     INFO - TEST-UNEXPECTED-FAIL | accessible/tests/browser/e10s/browser_caching_value.js | Test timed out - 
[task 2022-08-09T13:43:29.977Z] 13:43:29     INFO - GECKO(1152) | MEMORY STAT | vsize 1053MB | vsizeMaxContiguous 451MB | residentFast 278MB | heapAllocated 220MB
[task 2022-08-09T13:43:29.978Z] 13:43:29     INFO - TEST-OK | accessible/tests/browser/e10s/browser_caching_value.js | took 45354ms
[task 2022-08-09T13:43:29.989Z] 13:43:29     INFO - checking window state
[task 2022-08-09T13:43:30.000Z] 13:43:29     INFO - TEST-START | accessible/tests/browser/e10s/browser_events_caretmove.js

Hi James! Can you please take a look at this?
Thank you!

This seems to be a worsening of bug 1708336: same symptoms, just more frequent. I still have no idea what's causing this, nor can I reproduce it locally. I'm pushing a bunch more debugging to try to see if I can track this down, but I really don't hold out much hope.

To make things more chaotic, according to the screenshots, there seem to be a couple of different manifestations of this. In one, the test case doesn't even seem to load. In the other, the test loads, but a11y doesn't seem to fire the right events.

[Removed accidental duplicate comment.]

I've made some progress with my investigation, but it's slow work. I'm posting updates over in bug 1708336.

Raising to s3 because even though this is only on CI, it does severely affect the reliability of our tests on Windows and causes a lot of noise for all Mozilla devs. That said, hopefully this will soon be academic.

I'm hoping bug 1786496 (now landed) will fix this, but we'll probably need to wait another week or so to be sure.

No failures here for more than 30 days: https://treeherder.mozilla.org/intermittent-failures/bugdetails?startday=2023-01-19&endday=2023-02-18&tree=trunk&failurehash=all&bug=1783890