Closed Bug 1784931 Opened 2 years ago Closed 1 year ago

Intermittent dom/animation/test/mozilla/test_restyles.html | single tracking bug

Categories

(Core :: DOM: Animation, defect, P3)

defect

Tracking

()

RESOLVED FIXED

People

(Reporter: jmaher, Assigned: boris)

References

Details

(Keywords: intermittent-failure, intermittent-testcase, Whiteboard: [stockwell disabled])

Attachments

(2 files)

No description provided.

Additional information about this bug failures and frequency patterns can be found by running: ./mach test-info failure-report --bug 1784931

Update:

There have been 70 failures within the last 7 days:

  • 19 failures on Android 7.0 x86-64 WebRender debug
  • 1 failure on Linux 18.04 x64 WebRender asan opt
  • 1 failure on Linux 18.04 x64 CCov WebRender opt
  • 40 failuers on Linux 18.04 x64 WebRender tsan opt

Recent failure log: https://treeherder.mozilla.org/logviewer?job_id=387540770&repo=autoland&lineNumber=1975

Nicolas, any chance the frequency could have been caused by Bug 1777249? Looking at Orange factor
Thank you.

Flags: needinfo?(nchevobbe)
Whiteboard: [stockwell needswork:owner]
Assignee: nobody → csabou
Status: NEW → ASSIGNED

(In reply to Natalia Csoregi [:nataliaCs] from comment #4)

Update:

There have been 70 failures within the last 7 days:

  • 19 failures on Android 7.0 x86-64 WebRender debug
  • 1 failure on Linux 18.04 x64 WebRender asan opt
  • 1 failure on Linux 18.04 x64 CCov WebRender opt
  • 40 failuers on Linux 18.04 x64 WebRender tsan opt

Recent failure log: https://treeherder.mozilla.org/logviewer?job_id=387540770&repo=autoland&lineNumber=1975

Nicolas, any chance the frequency could have been caused by Bug 1777249? Looking at Orange factor
Thank you.

Bug 1777249 is a CSS-only change, used in DevTools, which isn't involved at all in test_restyles.html , so it's very unlikely to have been caused by that

Flags: needinfo?(nchevobbe)
Pushed by csabou@mozilla.com: https://hg.mozilla.org/integration/autoland/rev/6e76c0605c1e Disable mozilla/test_restyles.html on linux tsan and android debug for frequent failures. r=intermittent-reviewers,jmaher DONTBUILD
Assignee: csabou → nobody
Status: ASSIGNED → NEW
Keywords: leave-open
Whiteboard: [stockwell disable-recommended] → [stockwell disabled]

Boris, any chance you could help us assign this to someone?
Thank you.

Flags: needinfo?(boris.chiou)

I assign this to myself. Thanks for disabling this for frequent failures.

Flags: needinfo?(boris.chiou)
Assignee: nobody → boris.chiou

To a first approximation, ~all of the failures in documented history[1] are the following:

dom/animation/test/mozilla/test_restyles.html | CSS animations running on the main-thread should update style on the main thread on ::-moz-progress-bar - got 4, expected 5

That seems to correspond to this piece of code in the test's helper-iframe:
https://searchfox.org/mozilla-central/rev/e4afef5d3ff67781dc1377912344694f3cf3a226/dom/animation/test/mozilla/file_restyles.html#176,184,186-189

add_task(async function restyling_for_main_thread_animations_progress_bar_pseudo() {
...
  [markers, counter] = await observeStyling(5);
...
  is(markers.length, 5,
     'CSS animations running on the main-thread should update style ' +
     'on the main thread on ::-moz-progress-bar');
  progress.classList.add("stop");

This piece of the test seems to have been added in bug 1742393; adding dependency there. emilio, you added this test-snippet in question -- do you have any ideas about what might be going on that would result in 4 restyles when we expect 5 here, or ideas for diagnostics we might add?

(Also, if this proves elusive, we could conceivably add a paper-over-type-fix to the is(...) expression in question, to also allow a markers.length or 4 with todo_is in that case or something, just to stop the stem of intermittents if we don't have a way forward or cycles to investigate.)

[1] (327 of 338 failures since 2023-06-26 which is as far back as we have orangefactor data)

Severity: -- → S3
Depends on: 1742393
Flags: needinfo?(emilio)

(Also, if this proves elusive, we could conceivably add a paper-over-type-fix to the is(...) expression in question, to also allow a markers.length or 4 with todo_is in that case or something, just to stop the stem of intermittents if we don't have a way forward or cycles to investigate.)

Yeah I don't have any immediate idea about this failure, so this seems reasonable.

Flags: needinfo?(emilio)
Pushed by ealvarez@mozilla.com: https://hg.mozilla.org/integration/autoland/rev/7a9cbbb9f00d Allow 4 restyles in progress pseudo test. r=hiro
Status: NEW → RESOLVED
Closed: 1 year ago
Resolution: --- → FIXED
See Also: → 1869154
You need to log in before you can comment on or make changes to this bug.

Attachment

General

Created:
Updated:
Size: