mochitests can timeout in test-verify because of delays from MOZ_CHAOSMODE=0xfb
Categories
(Testing :: General, defect)
Tracking
(firefox113 fixed)
| Tracking | Status | |
|---|---|---|
| firefox113 | --- | fixed |
People
(Reporter: dholbert, Assigned: jdescottes)
References
(Blocks 3 open bugs)
Details
(Whiteboard: [no-nag])
Attachments
(1 file)
Filing this bug for the same core issue as bug 1734020, but now for mochitests. See bug 1734020 for more info; essentially, test-verify's chaos-mode runs have quite a bit of overhead (like orders of magnitude increase in duration in some cases); this makes some tests overshoot the threshold and trigger sporadic failures in the test-verify run.
We could of course paper over these on a case by case basis by sprinkling requestLongerTimeout all over the place, but that's kind of like whack-a-mole. It'd be better to increase the mochitest harness's timeout threshold for the chaosmode portion of the test-verify run, if we can do so.
Anyway: for now, we can use this bug as the dupe target for bugs of the form....
Intermittent TV [tier 2] .../tests/.../some-test.{js,html}
...where the failure occurs during the chaosmode portion of the test-verify run, and the failure looks like...
TEST-UNEXPECTED-FAIL | [path/to/test] | This test exceeded the timeout threshold. It should be rewritten or split up. If that's not possible, use requestLongerTimeout(N), but only as a last resort.
|
||
Comment 3•2 years ago
|
||
The following field has been copied from a duplicate bug:
| Field | Value | Source |
|---|---|---|
| Regressed by | bug 1811970 | bug 1812220 |
For more information, please visit auto_nag documentation.
|
|
||
Comment 4•2 years ago
|
||
This bug has been marked as a regression. Setting status flag for Nightly to affected.
|
Reporter | |
Comment 5•2 years ago
|
||
This bug isn't actually a regression, nor are its dupes. (They just may be flagged as such since they look like regressions in the treeherder timeline, with an orange job suddenly appearing for a particular test that was just touched. But in most cases the test would already have been just-as-orange if we ran it under test-verify on earlier builds, too.)
Adding [no-nag] to hopefully prevent further regression-related flag-twiddling from Release mgmt bot.
| Comment hidden (Intermittent Failures Robot) |
| Comment hidden (Intermittent Failures Robot) |
|
||
Updated•2 years ago
|
|
Assignee | |
Comment 8•2 years ago
|
||
FWIW, bug 1734020 actually improves the situation for browser mochitests as well by extending the overall timeout for the whole suite, but the issue is that the individual test timeout remains at the default value for the platform (eg 45s for opt).
|
|
Assignee | |
Comment 9•2 years ago
|
||
Bug 1734020 already extended the overall timeout for the test suite in verify mode, but individual tests remained capped at their initial timeout value.
This changeset aims to multiply the timeout by 2 for individual tests when MOZ_CHAOSMODE=0xfb is set in the environment variables.
|
||
Updated•2 years ago
|
|
||
Comment 11•2 years ago
|
||
|
||
Comment 12•2 years ago
|
||
| bugherder | ||
| Comment hidden (Intermittent Failures Robot) |
| Comment hidden (Intermittent Failures Robot) |
Description
•