Closed Bug 1859239 Opened 1 year ago Closed 1 year ago

[meta] Intermittent failures for remote tests when run on Linux with Wayland enabled

Categories

(Remote Protocol :: CDP, defect)

Product:
Remote Protocol
Component:
CDP
Type:
defect

Tracking

(firefox-esr115 unaffected, firefox118 unaffected, firefox119 unaffected, firefox120 wontfix, firefox121 affected)

Status:
RESOLVED FIXED
Tracking Flags:
Tracking Status
firefox-esr115 --- unaffected
firefox118 --- unaffected
firefox119 --- unaffected
firefox120 --- wontfix
firefox121 --- affected

People

(Reporter: whimboo, Unassigned)

References

Details

(Keywords: meta)

It looks like that remote tests have been enabled on the new Linux workers that run with Wayland. This bug tracks all the newly seen test failures that have been seen for these jobs.

Set release status flags based on info from the regressing bug 1857319

:ahal, since you are the author of the regressor, bug 1857319, could you take a look? Also, could you set the severity field?

For more information, please visit BugBot documentation.

status-firefox118: --- → unaffected
status-firefox-esr115: --- → unaffected
Flags: needinfo?(ahal)

Please note that all the failures are low-frequent so far and nothing stands out.

Flags: needinfo?(ahal)
Depends on: 1859310
Depends on: 1859463
Depends on: 1859720
Depends on: 1859934
Depends on: 1792475
Depends on: 1794938
Depends on: 1775790

Set release status flags based on info from the regressing bug 1857319

status-firefox121: --- → affected
Depends on: 1861320
Depends on: 1861542
Depends on: 1861708
Depends on: 1861697

Andrew, we get quite a lot of failures across all the remote tests for CDP when tests are run on Wayland. I wonder if we should temporarily revert or not given that the amount of intermittent failures is still low.

Flags: needinfo?(ahal)

I guess it's an infra problem rather than a problem with the individual tests?

If not, I'd be inclined to let the sheriffs disable the most frequent failures as part of their regular triage. If so, my preference would be to leave the tasks running and skip only the problematic manifests (e.g in the DEFAULT section).

Flags: needinfo?(ahal)

Right now the failures are wide-spread across all the tests. So there is no chance to actually skip portions of them.

I would suggest that we continue watching and if we detect some more frequent issue I might try to create a testcase for reproduction. Sadly right now this is not really doable.

Depends on: 1861881
Depends on: 1861813
Depends on: 1861767
Depends on: 1861765
See Also: → 1861933
Blocks: 1857319
status-firefox120: affected → wontfix
Keywords: regression
No longer regressed by: 1857319
Depends on: 1863926
Depends on: 1864332
Depends on: 1864326
Depends on: 1864993
Depends on: 1864927
Depends on: 1866694
Depends on: 1864443
Depends on: 1867188
Depends on: 1870789
Depends on: 1871611
Depends on: 1871459
No longer depends on: 1776024
No longer depends on: 1781337
No longer depends on: 1864326
No longer depends on: 1792475
No longer depends on: 1775790
No longer depends on: 1735283

We stopped running the remote browser chrome tests on Wayland due to the unstable worker setup. Once we can re-enable these failures should not appear anymore. As such lets close this bug as resolved now.

Status: NEW → RESOLVED
Closed: 1 year ago
Resolution: --- → FIXED
You need to log in before you can comment on or make changes to this bug.